Dfa Mod $ 5 Binære Alternativer

Den andre Virt Next APP registrerer en handel investor vil få et øyeblikkelig varsel som vil fortelle dem nøyaktig hva og hvordan å handle. Alle valgmuligheter har en utløp som er satt av handelsmannen, og dette kan variere fra 30 sekunder til så lenge som ett år 365 dager Live binær handel med virtnext Online Trading Software I Sverige Vi har ikke prøvd den autopiloterte binære handelsprogramvaren fra Virtnext Hvis du har brukt denne programvaren til å handle binære alternativer for deg, vær så snill å forlate en Hvis alternativet beveger seg i retning velg av næringsdrivende ved utløpet vil han eller hun vinne en gevinsterutbetaling på 89 -98 Vincent Bollore, CEO-utviklerne, har utviklet hele The Virt Next-systemet fra grunnen opp med målet om å utvikle et automatisert Virt Next-vintersystem Dette er en viktig drivende faktor bak buzz av Virt Next programvare og hundrevis av nye handelsfolk som fortsetter å delta i Virt Next daglig. Virt Virtes programvare kommer til å finne den beste trading muligheten det vil gi deg muligheten til å tjene opptil 97 profitt per vellykket handel. Det er flere investorer som når over 1,997 per dag ved hjelp av live trading-signaler fra Virt Next programvare. Hva kommer det til å ta for å komme i gang med VirtNext Software VirtNext programvare, begynn å motta live trading signaler og plasser din første binære Live binær handel med virtnext Azione Unicredit Milano Finanza Forex Disse nettstedene er sterkt anbefalt i binærvalgsbransjen for deres ærlige og upartiske tilnærming til all binær handel med Virtnext 4days ago Binær alternativer Handel med VirtNext er valutamarkedet og hvor kjøp og salg av valutaer på en som kan påvirke sin handel Virt Virt Next-programvaren aksepterer for tiden bare en rekke binære opsjonsmeglere, avhengig av hvor handelsmenn bor rundt om i verden. Vi har t prøvde autopilotert binær handelsprogramvare fra Virtnext Hvis du har brukt denne programvaren til å handle binære alternativer for deg, pl let leave a Det er et begrenset antall investeringsalternativer når det gjelder binær og det er samtalen eller putten The Virt Next-programvaren forteller handelsmannen å ringe alternativet skal bevege seg oppover av Signaler for utløp mens putten betyr Alternativet vil mest sannsynlig bevege seg nedover. Levende binær handel med virtnext Hdfc Credila Forex Card Credit. En tapt handel vil koste den første investeringen som ble plassert i det bestemte alternativet, som vanligvis varierer fra 5 til 1 K. En handel Live binær handel med virtnext. liste over aksepterte meglere er tilgjengelig etter at næringsdrivende går inn på deres detaljer på den offisielle Virt Next-nettsiden. Enhver som er interessert i handel med binære alternativer, trenger en konto med en akseptert binær opsjonshandel for å tjene penger på nettet uten investering i Wallis og Futuna. Disse nettstedene anbefales sterkt i binærvalgsbransjen for deres ærlige og upartiske tilnærming til all binær handel med Virtnext 4days ago Hva skjer det? Komme i gang med VirtNext-programvare På den tiden kan du laste ned VirtNext-programvaren, begynne å motta live binær handelstempler. Trade Options Vi har ikke prøvd autopiloted binær handelsprogramvare fra Virtnext. Hvis du har brukt denne programvaren til å handle binære alternativer for deg, vennligst legg igjen en Enhver ny investor som ønsker å komme i gang med programvaren, må investere med en akseptert binær opsjonshandelmegler. Produktnavn nå for tiden vellykkede binære alternativer, handelskvoter som er uovertruffen av andre binære valghandelssystem 2016 som er tilgjengelige, og det er s alt takket være den sofistikerte koden som driver hele Virt Next-systemet. Hundrevis av nye binærhandlere har flommet inn i Virt Next-medlemsområdet på grunn av det vellykkede handelsforholdet som Virt Next-programvaren nå når, og Vincent Bollore hevder at Virt Next-systemet har kraft og har vist seg å nå så høyt som 97 vellykkede handler for flere av våre beta-testere. Det er fortsatt en Bi g spørsmålet om hvordan Virt Next systemet opererer og faktisk gir en slik høy suksess når man investerer i binærvalgsindustrien. Den ble utviklet med brukeren i tankene og er i stand til å fullstendig ta kontroll over den binære opsjonshandelsprosessen ved å finne og automatisere handler for investor Nye binære opsjonshandler kan plugge inn live Virt Neste binære handelssignaler og begynne å motta sanntidsvarsler øyeblikkelig, slik at de ikke må bruke Tid å lære inn og ut av binær investering Live binær handel med virtnext Free Commodity Trading Systems Og metoder Den nyeste funksjonen som er lagt til Virt Next-programvaren, er automatiseringsfunksjonen som vil fullstendig kontrollere handel ved å plassere handler på Virt Next-brokeren. Live binær handel med virtnext Poenget bak Virt Virt-programvaren er å ta ut gjett arbeid og læringskurve til handel binære alternativer Ta for eksempel en handelsmann X gjør en handel på eiendelen Selv om du ikke har ha Har du noen erfaring innen binær handel, trading med Virtnext binær Hva er det som kommer til å ta for å komme i gang med Virt Next Software Hva kommer det til å ta for å komme i gang med VirtNext Software På den tiden kan du laste ned VirtNext-programvaren, start mottar live binær handel Binær Options Trading med VirtNext er valutamarkedet og hvor kjøp og salg av valutaer på en som kan påvirke sin handel. Auto Trading Etter å ha gjort en detaljert undersøkelse kom vårt team til den konklusjonen at det er bedre binære roboter til handel med enn VirtNext. Dette er en viktig drivende faktor bak buzz av Virt Next-programvare og hundrevis av nye handelsfolk som fortsetter å delta i Virt Next daglig. Virt Virt-programvaren kommer til å finne de beste handelsmulighetene som vil gi du muligheten til å tjene opptil 97 profitt per vellykket handel Det er et begrenset antall investeringsalternativer når det gjelder binære og det er samtalen eller putten The Vi rt Neste programvare forteller forhandleren å ringe alternativet skal bevege seg oppover med Signaler ved utløp mens putten betyr at alternativet vil mest sannsynlig bevege seg nedover. Det er flere investorer som når over 1,997 per dag ved hjelp av de levende handelssignaler som tilbys ved Virt Next software. Clever ideer for å tjene penger på nettet Togo Bourse Nouvelles Fx Lite Binær Options Client Blog. Mais ds mardi, une runion de la Commission of Schools, som er en avtale til Togo, og er en av de mest kriminelle politikkene som tilbys av CEREMONIE DE POSE DE LA PREMIERE PIERRE DU ROCARE TOGO EN LANCEMENT DU CONCOURS 2017 DU PROGRAM DE BOURSE Fidle sa tradition, Universitetet i Lom organiserer en rekke vellykkede turer. Journes N 1 de la tlphonie mobil au Togo, Togocel vous foreslår ses tjenester de pour toutes les burser, en fakturering mens du er interessert i Robot Forex 2016 Profesional Real For Five Signs Forex Rsi Indikatorstrategi Hvor Å Trade Forex Eller Stock Market RSI jeg ndicator er begrenset mellom 0 og 100, noe som indikerer områder av Du er her Forex Learning Center Nivå Eksperter Trading Strategies Forex For denne strategien kan vi bruke en hvilken som helst tidsramme La oss undersøke 1-timers diagrammet over USD JPY Vi bruker følgende indikatorer en 5-periode eksponentiell gratis nedlasting Indikatorer, RSI Indikator for Metatrader 4 Alle indikatorer på Forex Strategier Ressurser er gratis Her er det en liste over nedlasting RSI Relative. Software APP og Virt Next Investments LTD har bare gått til binære handelsfolk av Vincent Bollore A vellykket binær opsjonshandlerekspert som ønsket å dele sine binære opsjonshandelstrategier og hemmeligheter i et automatisk handelssystem Live binær handel med virtnext Virt Virt Next megler er plattformen som lar deg plassere handler og før du kan plassere en eneste handel du trenger å åpne og finansiere din handelskonto med en megler som arbeider med Virt Next and Options Trading Software ved å følge de signalene som programvaren gir de handlende kan begynne å investere med liten eller ingen bakgrunn i binære alternativer. Binært alternativ 81 Autotrader gjennomgang Minimumsinvesteringen er vanligvis 250, men dette kan variere avhengig av hvilken megler valgt. Den andre du finansierer din Virt Next binære opsjonskonto, får du umiddelbar nedlasting lenke til Virt neste APP. Bal des Conscrits de Besse. På forhånd foreslår de vennene dine at de ikke lenger er i stand til å se på, så du kan ha en pause ved siden av det, Omvendt er det bare å se på DVD-en. Du kan også se på mer enn 30 min. SALLE POLYVALENTE de BESSE. Du kan også høre musikken fra Extremusic Disco-mobile og La Buvette-serien. Du kan også se på denne filmen. på sera en modus dcompression et på sera donc en GRANDE, trs GRANDE FORME Les mer.81 har bookmarked denne hendelsen. Din reise starter nå. Reserver passord. Opplæringsprogrammer med løsninger. Fundamentals of Sequential Programs. Fo undasjoner vi ønsker å forstå hvordan sekvensielle programmer fungerer Vi ønsker å vite hvordan vi får fra programmet vi skriver til programmet som kompilatoren kjører. Slike programmer ikke multi-threaded eller samtidige.11 oppgaver hver med flere deler. lang, mye arbeid - start tidlig. due onsdager klokken 17.00 på Marmoset. we kan bruke C, C eller Racket - du kan lese på nettsiden for ting å vite for å bruke disse språkene i kurset. Oppgaver 25. du trenger minst 50 for midtveis 2 endelig 3.Midterm dato tirsdag 3. mars 2015 4 30-6 20 pm. Vi skal bruke piazza. no obligatorisk lærebok. Hva skjer når du kompilerer og kjører et program. Hva er en kompiler. høy nivå språk - kompilator - nivå language. this er vanlig, men ikke alltid saken. more generelt. source språk - kompilator - mål language. source program - kompilator - målprogram tilsvarende, for noen definisjon av tilsvarende. Hvorfor trenger vi en kompilator - å hjelpe mennesker skrive kode. ønske for folk å forstå og program. lets du velger det riktige språket for jobben du vil gjøre. sikkerhet - kompilatoren hjelper deg, lar deg vite om du gjør dårlige ting. abstraksjon - kan skrive programmer uten å forstå hvordan det virker. Hvorfor trenger vi en kompilator - hvorfor kan ikke datamaskinen bare Kjør kildespråket direkte. Språket på lavnivå er maskinvareavhengig maskinavhengig - optimalisert for å kjøre godt på bestemt maskinvare - men kildekoden er maskinavhengig. En nærmere titt på kompilatoren. CS241-kompilatoren - ikke nødvendigvis sant for hver compiler. scanning normaliserer innspillet, for eksempel normaliserer whitespace. Eksempel kan være en hvis setningen får tegnet IF. De to første trinnene er den syntaktiske analysen. Eksempler på monteringskoden legger til 3 5 7 jr 31. notat assembler er en spesifikk type kompilator som oversetter mellom forsamlingskode og maskinkode. Vi kommer til å være kodende for hele prosessen. Binary number. Bit 0 eller 1 Abstraction of high low voltages or magnets. Byte 8-bits eg 11001001.There are 256 2 8 mulige byte. Ordmaskin spesifikk gruppering av bytes.4 eller 8 byte 32-bit eller 64-bit. Vi skal bruke et 32-biters ord - dette betyr at vi genererer og monterer monteringsnummer. Nibble 4 bits Half en byte. Giv en byte i datamaskinens minne, hva betyr det For eksempel 11001001.Det kan være et nummer Konvensjonelt i binær er det 2 0 2 3 2 6 2 7 1 8 64 128 201. Det er en usignert verdi Vent , hvordan representerer vi negative tall. Simple tilnærming sign-magnitude representation. Reserve fi første bit for å representere tegnet 0 betyr positiv, 1 betyr negativ - resten av bitene er størrelsen. 11001001 er et negativt tall med størrelsen 1001001 73 - så verdien vår er -73. For 8-bit kan den representere tall formular -127 til 127.0 er både 00000000 og 10000000 - to representasjoner for null må sammenlignes for NULL - ikke bra. andre ulemper - aritmetikk er vanskelig. Hvis vi må positive eller to negative verdier, gjør det normale tillegget for magntiudes og bruk det felles tegnet. Men hva hvis tegnene er forskjellige Trekk den minste verdien fra den større verdien og bruk tegnet på det større - altfor kompliserte Ikke godt. Negative tall - bedre tilnærming - 2s kompliment. Tolk nummeret som unsigned. if den første bit er null, da ferdig. else, trekke 2 n.111 7 - første bit er ikke 0 - trekke 2 3 - -1.101 5 - første bit er ikke 0 - trekke 2 3 - -3.1100100 fra før er nå -55.To få twos kompliment negasjonen av et n-bit tall, trekke tallet fra 2 n. lekk bitene 0-1, 1- 0 og legg til en. Binær - Decimal. For 8 biter gir dette -128 til 127.note -128 ingen negasjon som 128 har ingen representasjon. eg med 3 bits 100 - 011 1 100 vi får det samme - ingen representasjon for sin negasjon. Bare en null. Aritmetikk er klar. Aritmetikk er mod 2 n. Du kan bare legge dem til som du vanligvis ville. Basen 16 0 9, og AF-tilfellet spiller ingen rolle. Hver heksesiffre er 4 biter, så 11001001 abonnement 2 C9 abonnement 16.subscipt 2 betyr at det er s i base 2.omkoble binært til hex ved å ta 4 sifferbiter. eg 11001001 - bryte inn i 1100 1011 - 1100 4 8 12 C 1001 1 8 9 - C9. bruk 0x som hex-prefikset f. eks. 0xC9. Gi en byte hvordan kan vi fortelle hvilken tolkning er riktig unisnged, sign-magnitude, twos complement. We kan egentlig ikke vite Vi må huske vår hensikt når vi lagret byte. Men vent Vi ikke t vet selv om det er et nummer. Det kan være et tegn - avhenger av tegnkodingen du bruker. Vi vil anta en konvensjon ASCII amerikansk standardkode for informasjon Intercha nge. what representerer 11001001 represent. number unisnged, sign-magnitude, to s compliment. it kan være en character. bit representasjon - så vi vet ikke hvilken karakter det er - vi skal bruke ord fortsatt, men med første bit 0.what chracter er 01001001 I. kan være address. random flags. instruction, eller en del av en våre instruksjoner er 32-bit. Vi kan egentlig ikke vite at vi må huske vår hensikt når vi lagret byte. Maskin Språk - MIPS. Hva ser en instruksjon ut. Hvilke instruksjoner er der. Vi vil bruke MIPS forenklet 18 forskjellige 32-biters instruksjonstyper. Maskinvaren og datapatelen. Hjernen til datamaskinen. Control unit. Fetches og dekoder instruksjoner. Koordinerer iput og output. Dispatches til andre deler av datamaskinen for å bære dem output. ALU aritmetisk og logisk enhet. Responsible for matte, logiske operasjoner og sammenligninger. MIPS 32 Generelle formål register. 0 er alltid 0. 30 og 31 er spesielle etter konvensjon. Et eksempel på registeroperasjon. Legg inn innholdet i registre s og t, og lagre resultatet i d. represented av d - s t.32 registre, 5 bits per register 2 5,3 registre for en instruksjon 15 bits satt til side for registreringer 17 bitere leftover for å kode inn operasjonen. multiplikasjon gir 64 bit resultat - første 32 biter er i HI og andre 32 biter er i LO. HI og LO butikkresultater, men du kan ikke skrive til dem direkte - det er ops å flytte verdier ut av høy og lav. Jeg kan også lagre resten fra divisjon og LO kvotient. RAM - Random Access Memory. Dette er hovedminne på datamaskinen. Dette er en stor mengde minne lagret vekk fra CPU. Dataene går mellom CPU og RAM og bussen. Vi tenker på bussen som 64 ledninger forbinder de to komponentene. RAM er bare et stort utvalg av n byte n.10 9 en Gigabyte. Hver byte har en adresse som går fra til n-1, men vi grupperer alt på ordens ordre , så vi vil bruke adresser delelig med 4, og hver 4-byte blokk er et ord. Words har adresse 0x0, 0x4, 0x8, 0xC, 0x10, 0x14, 0x18, 0x1C, 0x20 etc. Moving data mellom RAM og CPU. Transfer et ord fra en spesifisert adresse til et spesifisert register. Den ønskede adressen går inn i MAR-minnet adresseregister, deretter går ut på bussen. Når den kommer til RAM, de tilknyttede dataene sendes tilbake på bussen og går inn i MDR-minnedatabasen. Innholdet i MDR flyttes til destinasjonsregisteret. Lag akkurat som last, men i omvendt. Hvordan vet datamaskinen hvilken ord inneholder instruksjoner og som inneholder data overraskelse det gjør ikke det. Det er et spesielt register kalt PC program teller som holder adressen til NEXT instruksjon å kjøre. Med konvensjon, garanterer vi at noen faste adresser sier 0 inneholder kode og deretter initialiserer PC til 0.Then kontrollenheten kjører hente-utføringssyklusen. Hvordan blir et program kjørt. Det er et program som heter lasteren, som setter programmet i minnet og setter PC til adressen til den første instruksjonen i programmet. Hva skjer når programmet endene. Vi n eed å returnere kontroll til laster-PC er satt til adressen til neste instruksjon i loader. Which instruksjon er det. 31 vil alltid lagre den riktige adressen for å komme tilbake til, så vi trenger bare å sette PC til 31. Merk at bruk av 31 er konvensjon. Vi vil bruke kommandoen joppregister jr for å oppdatere verdien av PC. hardware architecture. machine språk. main minne vs registers.32 generelle formål registre. 31 returadresse - på programstart for å returnere adressen til lasteren, bør programmene våre alltid avslutte ved å hoppe til returadressen 31 ved hjelp av kommandoen hoppregistrering jr. 30 vet vi ikke ennå. PC - programteller - holder adressen til neste isntruksjon for å utføre. Eksempel legg til 2 verdier i registre 5 og 7, lagre resultatet i register 3, og deretter tilbake. Eksempel legg til 42 og 52, butikk sum i 3 og returner resultatet i register 3, så return. xxd - cols 4.this kommandoen vil vise hex code. xxd - cols 4-bits. adding bits-alternativet viser binary. the filstørrelse er instruksjoner 4 bytes. We vil begynn å skrive våre programmer ikke i binær og hex, men med enkle mnemonics. There er en oversettelse tilbake til den nødvendige binære assembler Hver monteringsinstruksjon tilsvarer en maskin instruksjon nesten. Vi vil gå tilbake til det forrige example. beq gå et annet sted hvis to registre er equal. bne gå et annet sted hvis to registre ikke er like. both instruksjon øke PCen med et gitt antall ord fremover eller bakover. Hva er verdien av PC før avdelingen blir utført. Basert på hente-utfør syklus, har PC allerede blitt økt for å peke på neste instruksjon b før instruksjonen er avkodd og utført. Derfor er offset i forhold til neste instruksjon. En annen kommandoslutt - sett mindre enn. a -1 hvis bc, et sett til 0 ellers. Eksempel beregne den absolutte verdien av 1, lagre den i 1 og return. To gjøre dette, må vi bruke grener og hopper for å endre PC. how om vi starter med c-versjonen hvis x 0 x - x. Eksempel sum totalene 1 13, lagre i 3 og returner c. Hva om vi ikke kan bruke loopkonstruksjoner - goto. Branch offset -5 fordi PC peker på neste instruksjon. Effektivitet i kodeinnstilling 1 til 1 på hver iterasjon av loop. Last Time Assembly Programming. sum heltall 1 13, lagre i 3 og returnere. Programmer med RAM. lw er belastningsordet instruksjon. laster et ord fra RAM til et register. syntax lw a, jeg b. loads ordet på MEM b i til a. b er base register, jeg er offset. sw er butikken ord instruction. stores et ord fra register til RAM. syntax sw a, jeg b. stores ordet ved a i minnestedet MEM b i. Exempel 1 holder adressen til en matrise og 2 holder lengden av matrisen Hent elementet med indeks 5, og lagre det i register 3.Nå gjør vi det igjen, men mer abstrakt. I dette tilfellet er indeksen ikke kjent. Vi trenger. mult multiplikasjonsinstruksjonen. Som å multiplisere to 32-biters tall kan det resultere i et 64-biters tall, blir resultatene lagret i to spesielle registre hei og lo. syntax mult a, b. div divide instruksjonen. Kvoten er strøket i lo, og resten er lagret i hi. syntax div a, b. mfhi og mflo flytte fra HI og flytte fra LO instructions. move verdiene fra hei eller lo henholdsvis til et gitt register. syntax mfhi d, mflo d. Etiketter reagerer på loop. recall sløyfen vi hadde Sum integlene 1 13, lagre i 3, deretter return. the lis kommandoen var innenfor vår sløyfe, som kunne flyttes utenfor. Dette er greit, men nå har bne på slutten en feilaktig øyeblikkelig bør være -3.I nesterløkker er dette et mareritt Dette skyldes at vi brukte eksplisitt forgrening, ved å legge til fjerning av instruksjoner betyr det at vi må endre grenseoverskridelser. I stedet tillater assembler merket instructions. syntax er etikettinstruksjon. navnetikett med instruksjonen. Når montøren ser en etikett, beregner den forskjellen mellom PC og topp, i form av words. eg over etikett-PC 4 0x14 -0x20 4 -0xC 4 -12 4 -3. Nå kan vi omskrive loop-eksemplet med en etikett. Prosedyrer Prosedyrefunksjoner ved montering gjør at vi kan gjenbruke kode. Vi har to problemer å løse. Kall og retur Hvordan overfører vi kontroll inn og ut av prosedyren, parameterpassasje, etc. Registers Hvordan sørger vi for at prosedyrene ikke overskriver viktige data. Vi vil begynne med det andre problemet. Vi kan reservere registre for pro-prosedyren f, og noen for hovedlinjen uten overlapping, men hvis prosedyrer kaller andre prosedyrer eller seg selv, går vi tom for register. I stedet vil vi tillate prosedyrer å gjøre uansett hva de vil ha med registre, så lenge det setter dem tilbake til sine opprinnelige verdier ved utgang. Derfor må vi bruke RAM Hvordan holder vi prosedyrer fra å bruke samme RAM. For å forhindre at prosedyrer bruker de samme RAM-adressene, bruker vi en Stabpekeren Stabepekeren er plassert i register 30, som er det andre begrensede registeret som vi tidligere diskuterte. Vi tildeler vanligvis minne fra toppen eller bunnen av ledig RAM, og på en eller annen måte holder rede på hvilket RAM som ikke er brukt. maskinen hjelper oss ut 30 er initialisert av senderen til bare forbi det siste minneordet. Vi bruker 30 som et bokmerke for å skille brukt og ubrukt RAM, hvis vi allokerer fra bunnen. Vi bruker RAM som en stabel, går opp på funksjonen samtaler og ned på returns. Our strategi skyver hver prosedyre registre det vil bruke på stabelen og popper opprinnelsesverdiene fra stabelen når den er ferdig 30, stakkpekeren inneholder adressen til toppen av stakken. Templat for prosedyrer. Eksempel Anta at en funksjon f bare bruker 2 og 3.Hvordan lagrer vi registrene og justerer deretter stavelpekeren. Fordi en gang registerene er lagret, kan vi bruke en til å holde stakkjusteringen vlaue. Assembly programmeringsprosedyrer. Maskinen har bare 32 registre Hvordan kontrollerer vi at viktige data ikke er overskrevet. Kode for å ringe og returnere. Når vi kommer tilbake fra en prosedyre, må vi sette PC til linjen etter jr HER Hvordan vet vi hvilken adresse som er. jalr-kommandoen, er hopp og lenkeregister. Undervisningen er akkurat som jr, men det setter også 31 til adressen til neste instruksjon, dvs PC. assembly kode ser nå ut. For å komme tilbake til lasteren må vi lagre 31 på stakken først og deretter pope den før retur. Det enkleste alternativet er å sende retur parametere via register. Men dette gjør det komplisert med hensyn til å finne ut hvor register parameters er. Hvis dette er gjort, må prosedyreskriveren dokumentere sin kode slik at klienten vet hvilke register som skal sendes bakover og fremover. Det største problemet er at det bare er 32 registre Et bedre alternativ skyver parametere på stakken Dette krever også dokumentasjon. Fullkodeeksempel - Vi vil skrive en funksjon som vil summere de første n tallene. Ingen ekstra maskiner er nødvendig. Hvis du registrerer, parametere og stakkenadministreres riktig, vil rekursjonen fungere. Input og Output. Input støttes ikke. Deal with it. Output MIPS gir en plassering 0xffff000c kalt videominne, for å lagre ord hvor den minste signifikante byten vil bli skrevet ut til skjermen. exempel utskrift CS, etterfulgt av ny linje. Assembler A Assembler er et program som oversetter forsamlingskoden til ekvivalent maskinkode. Sosial kode legger til 1, 0, 0 - assembler - maskinkode 0100010. Enhver oversettelsesprosess innebærer to trinn. Analyse forstår hva som menes med kilde string. synthesis output tilsvarende målstreng. En samlingsfil er bare en strøm av tegn, en tekstfil. Ta 1 gruppe tegn inn i meningsfulle tokens. For eksempel etiketter, heks nummer, vanlig talord, registre etc. Denne delen har blitt gjort for oss, vi skal snakke om det i mer detaljert. f. eks. for C-startkoden, er hver token en forekomst av token-klassen. Steg 2 Gruppens tokens i instruksjoner, om mulig analyse. Steg 3 Output-ekvivalent maskinkodesyntese. Hvis tokens ikke er ordnet i fornuftige instruksjoner, sendes FEIL til standardfeil. Advice det er mange flere feil konfigurasjoner enn de rette for en forsamlingsfil. For eksempel vil instruktøren beq 1, 0, abc generere en sekvens av token-typer ID REGISTER COMMA REGISTER COMMA ID. Største problem med skrive assembler Hvordan samler vi. Vi kan ikke sette sammen dette fordi vi ikke vet hva abc er. Løsningen vi skal skanne gjennom programmet to ganger. Pass 1 gruppere tokens i instruksjoner og ta opp adressene til alle merkede insturctions - et symboltabell som er konseptuelt en liste over navn, adressepar. En samlingslinje kan ha mer enn en etikett. Passord 2 Oversett hver inntrengning til maskinkode Hvis en instruksjon refererer til en etikett, kan du slå opp den tilknyttede adressen i symboltabellen. Vi skal samle inn den monterte MIPS til stdout, og vi bør ouput symboltabellen til stderr. jalr for call. jr for return. Assembler 2 passerer forsamlingskode - lexer gitt til oss - tokens - pass 1 - symboltabell mellomliggende representasjon - pass 2 - maskinkode. To passerer Pass 1.group tokens i instruksjoner og bygge symboltabell se på forsamlingssporet over fra slutten av siste klasse. For å finne adressene, er det enklest hvis du skriver dem inn ved siden av hver linje av assembly. translate hver instruksjon. 2 - 0x00000104 13 - 0x0000000d. look øverst i symboltabell. kalkuler topp-PC 4 0xC-0x20 4 12-32 4 -20 4 -5.Vi får 0x1440fffb. to negere en to s kompliment verdi, flip bitene og legg til one. To samle bne 2, 0, top top - -5, ser vi på 0001 01ss ssst tttt iiii iiii iiii. opcode 000101 5.first reg 2 00010.sekund reg 0 0000.Problem vi trenger å montere 6 bit opcode , de to 5 bitene av regitere og de 16 bitene med umiddelbar verdi i en 32-bit instruksjon. Hvordan signert int instr. racket arithmetic-shift 5 -26.problem -5 0xfffffffb, vi har bare plass til 16 bits. bit - wise AND two 1s gir 1, noe annet gir 0. når vi bitwise - og med en 1, er inngangen uendret når vi bitwise - og med 0 er resultatet 0 Således bruker vi det til å slå av bitene. når vi bitwise - eller med en 1, er resultatet 1 når vi bitwise - eller med en 0 inngangen er uendret. Vi bruker det til å slå biter på. For å fikse -5, kan vi gjøre bitvis og med 0xffff. -5 0xffff 0xfffb. Dette slår av brikkene mer signifikant enn vi trenger bare å holde de siste 16 bitene. Dessverre sender vi 339804155, som er desimalrepresentasjonen og er 72-biters. Dette er dårlig. Når vi skriver ut en int, figurerer den ut sifre og deretter skriver dem ut Hvis vi skriver ut et kar, skriver vi ut ASCII-koden nøyaktig. Dessverre er det bare 8 biter, vi skriver ut fb av 1440ffffb som ønsket, så vi konverterer instruksjoner til 4 chars. in racket bruker skrivebyte og , for eksempel bitvis og -5 xfff. Det er mulig å lage en løsning ved hjelp av fagforeninger og bitfelt. Det er ikke så bærbart som forrige løsning, men veldig elegant. Feilsøking - inkludere strengen Feil et eller annet sted i feilmeldingene i rackett - Feilfeil. Loader La oss starte med å skrive et operativsystem. stjernen er i hovedsak en laster og gjør dette. 0S er også et program Hvor sitter det i minnet Andre progamer kan også løpe Hvor er de i minnet Og mer generelt er det kan være annen kode i minnet, for eksempel biblioteker. Vi kan velge en annen startadresse for programmer ved montering. Men da vet hvordan lasteren vet hvor de skal lastes. Hva om to adresser er de samme. Vi kan ikke laste inn programmer hvor som helst Etiketter kan løses til feil adresse eller bne beq. Lasten må på en eller annen måte fikse id --- vi legger til. programmet skal lastes på. id representerer en absolutt adresse. ord konstant dette er greit, vi trenger ikke å fikse det. Hva om grener med et id - nei, fordi montøren kalkulerer en relativ kompensasjon basert på PC, så flytting er unødvendig. Tips for Marmoset. Refer til MIPS forsamlingsreferanse shet. User spesifikasjon for å lage dine egne tester. Kan brukes til å sammenligne mot - din assembler shoud gir samme output as. Loader - vi kan ikke anta at vår kode alltid vil bli lastet på en fast adresse f. eks. 0x00 Hvorfor. Det kan være andre programmer eller kode i minnet. Vi har antatt at koden vår alltid er lastet på 0x00 Hvis denne antagelsen ikke lenger holder, hva bryter Labels. Hvilke instruksjoner må stilles. Lasteren må på en eller annen måte fikse. ID må legge til en konstant ID referanser en absolutt adresse. Consatnt Flytter IKKE. Hva om Brancher. no, fordi Assembler beregner en relativ kompensasjon basert på PC, så flytting er unødvendig. Alt annet inkludert BNE, BEQ Flytter ikke. Nå har vi OS 3 0.Problem igjen - dette vil ikke work. Assembled fil er en strøm av biter - hvordan vet vi hvilken som kom fra en med et ID og som er instruktører? Vi kan ikke Vi trenger hjelp Vi trenger mer info fra assembler. We merker at produksjonen fra de fleste montører ikke er ren maskinkode det kalles objektkode En objektfil inneholder binær kode, men inneholder i tillegg tilleggsinformasjon om filen som vil være nødvendig. Vi bruker objektfilformell MERL MIPS Executable Relocatable Linkable. Merk at dette er laget for CS241, så prøv ikke å undersøke den. Innenfor en objekts fil. koden i binær. som linjeledere må flyttes fordi de er id-instruksjoner. Andre ting som skal legges til senere. Headers har en informasjonskapsel for å fortelle oss at det er en MIPS-fil, og deretter lengdene på t han er helt fyldig og koden. Formatkoden er alltid 1 for flyttingen, og den tilhørende adressen er adressen i MIPS-koden til det flyttbare ordet. Header er alltid av størrelse 12 I vårt eksempel på lysbildet er koden størrelse 32 og symboltabellen er størrelse 16.Vi merker at 0x10000002 er MIPS for beq 0, 0, 2, det vil si en kommando for å hoppe over header, slik at MERL-filer kan utføres som vanlige MIPS-programmer hvis de lastes på 0x00.Vi vil også at montøren å generere flyttbar objektkode som vi gjorde ved å legge til relocs i eksemplet som er som vi vil få til slutt. Relocation Tool. input merl fil og en flyttingsadresse. output ikke-flyttbar mips-fil med merl header og footer fjernet, klar til å lastes på adresse. Lader Tools. both av disse ta et valgfritt 2. argument, som er adressen for å laste mips filen. Eksempelbelastning på 0x10000.we notatflytting gjøres vanligvis av lasteren - vi gjør det på denne måten som en visualisering av prosessen. Loader Relocation Algorithm. What vi kom opp med Sist klassen virker generelt, men kan fortsatt bli ødelagt. Hvis vi hopper til en consatnt adresse eller gjør matte på en etikett, er vi i trøbbel. Dette er imidlertid dårlig praksis. Dette er et skifte i rollen som etiketter da de ble innført var de en convenience, but now they are a necessity Essentially, if you want relocatable code, use labels. We often find it convenient to split MIPS programs into smaller units, for the same reason as with higher-level languages There is a major issue How can an assembler resolve a reference to a label in a different file. We have and with shared labels. Solution to assemble, concatenate the files and then assemble. cat java. This works, but it doesn t save us any work if I make a small change in but is huge. Also, I may only want to give someone the binary not the source We want to compile first then link after. Solution 2 Can we assemble first and then cat. header binary reloc table. No, they are all assembled to start at address 0, but the second two would need to relocatable. Thus w e need to assemble to MERL, not just MIPS However, if we cat two MERL files, we don t get a MERL file. Solution we need a tool smarter than cat. It must be able to understand MERL files and put them together intelligently. Such a tool is called a linker. And yet what should an assembler do with references to lables it cannot find. We will need to change the assembler. Aside what happens conceptually when I run g. asm o kind o lexer o -- assembler outpiler generates asm o, kind o, lexer o. o extention used to represent object files. linker combines the o files into an executable. Import and Export. We lose a valuable error-check. If a label is not defined, then our old assembler assumes an error. but now we just expect to find it within a linked-in file. How can an assembler tell between errors and intentional behaviour. The new assembler directive. id tells the assembler to ask for id to be linked in This does not assemble to a word MIPS. if we have a label abc, and that label does not occur, and no abc, then we have an error. Format code 0x11 means External Symbol Reference ESR. we need the name of the symbol, the address of use where the blank needs to be filled in. ESR entry contains. word 2 location of where symbol is used. word 3 length of symbol in chars n. word 4- 3 n ASCII chars in the symbol name one word per char. example ESR entry. How do we know which vsion of abc to use We can t assume labels won t be duplicated. makes id available for linking - this does not assemble to a word of MIPS. tells the assembler to make an entry in the MERL table. MERL entry tpe External Symbol Defintion ESD. ESD entry contains. word 2 address the symbol represents. word 3 length of symbol name n. word 4-3 n ASCII name. example ESD Entry. MERL now contains code, relocator addresses, addresses and names of ESR and EDR entries. Note that instructions referring to external symbols must also be relocated. Augment MERL files to help linker. MERL Header Code Relocation and External Symbol Table. Symbol table contains relocatable addresses, adresses and names of ESR imports and ESD exports. The linker Given two MERl files - how do we link them What are the steps. a - we want to subtract the header from m2 and then push forward by length of m1 code a relocate by a add a to every address in. resolve symbols and check for duplicates in exports - ERROR. if AND 0 then ERRROR for all addr1, label in do if exists addr2, label in then - addr2 remove addr1, label from add addr1 to for all addr2, label same. merge symbol tables. imports and exports and relocates and. output linked program with new header. output MERL cookie output totalCodeLen total imports, exports, relocates 12 output totalCodeLen 12 output output output imports, exports, relocates. Assembly language has a simple structure, is easy to parse, straightforward and unambiguous. Now we turn our attention to the compiler Source program - compiler - assembly code equivalent meaning. First step of our compiler source program - scanner lexical analysis - tokens. A high-level language is a more complex structure It is harder to recogn ize, and we have no single translation to machine code To convert to machine language we use a compiler. How does a compiler recognize if a given program is valid Note we re not asking if it s logically correct a much harder problem , but just whether the sequence of characters forms and allowed program based on the specifications of the language. Eventually, once the compiler can recognize whether a program is valid, how do we translate the program into an equivalent program in the target language usually lower-level. Is this a valid c program. what about the same program with a b 5 s. no it will be parsed as a b which is invalid. nope, same interpretation as above. How can we handle this complexity. We want a formal theory of string recognition - general principles that work in the context of any programming language. alphabet finite set of symbols e g typically denoted. String word finite sequence of symbols from a, abc, cbca, etc. Length w of symbols in w. Empty string is the empty string, not a symbol 0.Language a set of strings e g , b, aab, aaaab. is the empty word, while is the empty language. is a singleton language containing only the empty word. How can we automatically recognize whether a given string belongs to a given language This is mindblowing, and depends on how complex the language is Since the answer is dependent on language complexity, we will charactertize languages according to how hard recognition process is we will create classes of languages based on the difficulty of recognition This is based on Chomsky Hierarchy. Finite easy. Regular not so easy. Context-free pretty decent. Context-sensitive challenging. Recursive difficult. anything-else impossible. We want to study high-level languages at as easy a class level as possible, and move down when we have to. These languages have finitely many words. Write code to answer w L such that w is scanned exactly once, without storing previously seen characters. We can recognize a word by comparing each word in the finite set of words - but we won t use that approach. started compi ler. formal languages. chomsky hierarchy. context sensitive. very losely, recursive means that there is an algorithm or program to determine whether any given input is a member of the language. An abstraction of this program. Circles are states - configurations of the program based on input seen. The double circles are accepting if the program halts there. Example - mips operators. Since programming langagues don t usually admit only a finite many programs, finite languages are not much use. These languages are built from finite languages, but also support the following operations. Example show that is regular. This syntax is very tedious We have Regular Expressions. Question - is the language C regular. We have IDs a-zA-Z. A C program is a sequence of tokens, each of which come from a regular language. How can we define which sequences of tokens are valid We don t know yet. Thus, our answer is maybe. Can we recognize arbitrary languges automatically. Can we harness what we learned about recognizing fini te languages LOOPS. These machines are called Deterministic Finite Automata DFAs. Always a start state. For each character in the input follow the corresponding arc to the next state. If in an accepting state when input is exhausted, accept - else reject. What is missing. What if there is no transition Consider the ab example - what if our input is ab If we fall off the machine, reject More formally, an implicit error state exists all unlabelled transitions go there An error state has an all input loop back to itself and is non-accepting. Example String over with an even number of a s and an odd number of b s. Formal definition of DFA a DFA is a 5-tuple , Q, q0, A, , where. is a finite non empty set alphabet. Q is a finite, non-empty set states. q0 is an element of Q start state. A is a subset of Q accepting states end states. Qx - Q Transition function state and input symbol and gives next state. consumes a single character of input. we can extend to a function that consumes an entire word. Thus, a DFA, M , Q, q0, A, accepts a word w if q0, w is in A. example from before. Last time Regular Expressions. Handy to know precedence highest to lowest. Example a bc is a b c. A DFA, M , Q, q0, A, accepts a word w if q0, w is in A. If M is a DFA, we denote L M the language of M , the set of all words accepted by M L M. L is regular iff L L M for some other words, the regular languages are accepted by DFAs. Proof we will prove this later. Implementing a DFA. Unfortunatley this is very tedious Instead, we could use a lookup table see provided assembler starter code of states in columns and characters in rows - where they intersect is next state. Currently, our DFA takes an input string and returns yes no whether the given string is in the language However, it is possible to add an output facility to a DFA, resulting in a transducer. DFAs with actions can attach computation to the arcs of a DFA For example, consider L , where we compute the value of the number. sorry for horrible quality. another possible action would be emit a token - here we emit the current value of our number. Non-Deterministic Finite Automata What do we gain by making our DFAs more complex e g L w ends in abb. What if we allowed more than one arc edge with the same label from the same state. The machine choses one this is non-deterministic. We can think of it as magic, that it knows where it goes - or we can think of it as trying every path and see if there is any path that ends in an accepting path, then it accepts. We will accept if some set of choices leads to an accepting state Returning to our example.- start - a - - b - - b - start loops back to itself with a and b. The machine guesses to stay in the first state until it reaches the final abb, then transitions to accepting. NFAs are often simpler than DFAs. Formally NFA is a 5-tuple , Q, q0, A., Q are finite non-empty sets alphabet, states. q0 start and A subset of Q accepting. relation - Qx -- subsets of the powerset of Q 2 Q. This is the powerset of Q, which makes it non-deterministic The powerset is the set of all subsets of Q includes the empty set and Q. We want to accept if some path through the NFA lead to an accepting state, and reject if no such path exists. for NFAs sets of states x - sets of states. P, empty P, where P is a set of states. P, cw union of q, c for all q in P w. And we accept if, w union A 0. NFA Simulation Procedure. work with the following NFA, we will simulate baabb. BUT WAIT I am adding the right column to the table above as we speak If we give each set of states a name, and call those states, every NFA becomes a DFA For example. note if we draw this it looks just like our DFA from before. Holy cow, we have a DFA because D contains an accepting state. Consider L U with an even of a s We can draw an NFA. The DFA is much harder to produce Let s do a trace. Now we build the DFA via the subset construction. Accepting state are any states that includes an accepting state from the original NFA Every NFA hs an equivilent DFA, and NFAs recognize the same class of languages. What if we let ourselves change state without consuming a character We call these transitions we label the arrow as. This is a free pass to a new state without reading a character. This makes it easy to glue smaller automa together. Revisiting the abov e example. By the same renaming trick as before ie the subset construction every - NFA has an equivilent DFA Thus, - NFAs and DFAs recognize the same class of languages. Yes, DFAs and NFAs are finite state machines and the class of languages accepted by FSM are regular languages. Proof of Kleene s theorem one way. L is regular if L L M for some DFA M. If we can find an - NFA for every regular expression, then we have proved one direction of Kleene s theorem. The below pictures show empty languages, languages, single caracter, alternation, concatenation, repetiton. Thus every regular language has an equivalent NFA, which has an equivalent DFA and the conversion can be automated. Is C a regular language. C keywords, identifiers, literals, operators, comments, punctuation. These are all regular languages, and sequences of these are also regular, so we can use finite automata to do tokenization scanning, lexical analysis. Ordinary DFA can only tell us if a word is in a language We need something that takes an input string w, breaking w into w1,w2 wn such that each wi is in L and then output each wi. Consider L is regular. Let ML be the DFA that recoginizes L Then the second representation M which is a nonempty sequence of tokens is NFA modified from the NFA for ML - using t he rules from above for. We can add an action to each each - move such as output a token Our machine is non-deterministic - moves are optional. The question does the current setup guarantee a unique decomposition w w1 w2 2n The answer is no. Consider just the portion of the machine that does IDs.- --a-zA-Z-- --a-zA-Z0-9-- back to ending state. output token from ending state back to start state. The input abab could be interpreted as 1,2,3 or 4-tokens. How can we fix this We could always return the longest possible next token However, this can still fail. Consider L , w aaaa. If we take the longest possible token, we take aaa first and then we are stuck. However, we could have tokenized the string successfully as aa and aa. Oh well It would be bad to design a programming language like this. But wait many of the languages we use are designed like this. Remember the example when we started. c a b - a b NOT a b. Again from out example, c a b - a b which is invalid. But, a b would be valid - but that s not what the compiler does. Another example from c. vector vector v. c treats the like the bitshift operator instead of two closing angle brackets We have to trick the scanner. vector vector v. Fixed in c 11. Maximal Munch Algorithim. Simplified Maximal Munch as above, but if we are not in an accepting state, when no transition possible, then we ou tput an error don t backtrack. Our identifier must start and end with a letter and may contain a single. We also accept operator. For input ab. we find a valid start state, then b, and then -, but we cannot scan past this point since no further action is possible. Thus we would go back ab since it was the last valid point, and then continue from there. THe SMM algorithm would simply find an error but in practice this is usually sufficient. Note that if we have L we show it s regular by describing it with a regular expression - using only operations in regular expressions that we talked about - here it s aa. Scanning algorithm - Maximal Munch vs Simplified Maximal Munch. difference - no backtracking. Now consider alphabet of brackets. Can we build a DFA for L. Each new state lets us recognize one more level or nesting, but no finite number of states recognizes all levels of nesting, and a DFA must have a finite number of states. tokens Parsing syntactic analysis parse tree. what is required. Given a formal syntax specication, the parser verfiies whether the incoming tokens have a valid syntax confirming to the specifications and outputs a parse tree to allow semantic analysis in the following step. This cannot be done with regular languages, so we move to the next class in the Chomsky Hierarchy. These are languages which can be described by a context-free grammar a set of rewrite rules. Following the balanced paranthesis example. a word in the language surrounded by S - S. the concatenation of two words in the language S - SS. The shorthand for this is S - S SS. Example show how this system generates. The notaton we have used. means derives. a - means can be dervied from a by one application of a grammar rule. we formally define a context-free language to consist of. an alphabet of terminal symbols. a finite non-empty set N of of non-terminal symbols, such that N union null set. we often use V vocabulary to denote N U. a finite set of P productions, where productions have the form A - , A in N, in V. an element S in N as the start symbol. In our parentheses example, we have that terminal symbols are , non-terminal symbols are , and S is our start symbol. a, b, c are elements of characters, terminals. w, x, y are elements of words. A, B, C S are elements of N non-terminals. S is the start symbol., , indicate elements of V. We write A if there is a production A - in P. means that delata1 1 n for n 0.Definition L G , where L G is the langauge specified by G. Note that this is a set of strings of terminals only A language L is context-free if L L G , for some context-free grammar G. Example Palindromes over. S - a b c aSa bSb cSc. We can also define this as. S - aSa bSb cSc M. Show that S abcba. S aSa abSba abMba abcba. This proccess is called a derivation. But we could have speciiced this with a regular expression. Show that S a b. S S op S a Op S a S a b. alternatively S S op S S op b S b a b. Note that we have a choice over which symbmol to expand first. Leftmost Derivation always expand leftmost symbol first. Rightmost Derivation always expand rightmost symbol first. We are showing too much information, since this information doesn t matter and this can get very long We can express them naturally and succinctly as a tree stucture. Example consider abcba. For the leftmost derivation, there is a unique corresponding parse tree and vice versa. Midterm - we will provide. MIPS reference sheet. MERL file format spec. ASCII chart. CFL context free languages CFGs. In a derivation, one non-terminal in the current string is selected and replaced or rewritten using RHS of a production rule for that non-terminal. e g production rules. What is the RE for this langugae. Leftmost derivation for a bxc S S Op S a Op S a S a S Op S a b Op S a b x S a b x c. Or, expand the first S first S S Op S S Op S Op S a Op S Op S a S Op S a b Op S a b x S a b x c. These correspond to DIFFERENT parse trees A grammar for which the same word has more than one distinct parse tree is called ambiguous The grammar we defined above is ambiguous If we only care about whether a string is in the language, the ambiguity doesn t matter. As compiler writers we want to know why a word is in a language, so the derivation matters Why does it matter The shape of the parse tree describes the meaning of the word, so a word with an ambiguous parse may have multiple possible meanings. In the first tree, bxc is grouped more tightly, and in the second a b is grouped So it could mean a bxc or a b xc. END MIDTERM CONTENT. What do we do about this. use heuristics precedence to guide the derivation proccess. Make the grammar unambiguous. Then we have a strict left to right precedence. new producton rules that prevent uncessary brackets. new derivation NOTE use single arrows for production rules and double arrows for derivation. If L is context-free, is there always an unambiguous grammar. No There are ineheritently ambiguous languages that only have ambiguous grammars. Can we construct a tool to tell us if a grammar is ambiguous No This is undecidable There is provably no algorithm to solve this problem The equivalence of grammars is also undecidable. We need to build a recognizer - what class of programs recognize CFLs. regular languages are represented by DFAs, a program with finite memory. Context-free languages must recognize infinite memory, they are represented by NFA with a stack. But we need more than a yes no answer We don t want to have a program and the compiler just say yes or no. We ca n use grammars to specify the syntax of a language For example, a while loop. Parse trees allow us to understand the program We need to know the derivation parse tree and error diagnosis The problem of finding the derivation is called parsing How do we use grammar to go from source program to a parse tree. We have two choices. forwards - top down - start at S, work to w. backwards - bottom up - start at w, work to S. How can we derive S abywx We want an algorithm to generate the derivation. Consider a leftmost derivation. What are we doing here Match input symbols startng from left until you encounter a non-terminal Replace non-terminal with RHS of a rule and continue matching more formally. Use the stack to store s in reverse and match against characters in input. Invariant the consumed input plus the reverse of the stack contents is equal to i Sean calls these left right sentential forms - for leftmost rightmost derivations. For simplicity, we will use augmented grammars for parsing We invent two new symbols BOF and EOF and a new start symbol S. say w BOF a b y w x EOF. Let s codify this. When the top of other stack is a terminal pop and matched against the input. when TOS top of stack is a non-terminal A, pop A and push reverse , where A - is a grammar rule. accept when stack and input are empty. brute force try all combinations until one works - this is inefficient. we want a Deterministc Procedure with no backtracking. our solution is to use the next symbol of input lookahead to help decide. Construct a predictor table - given a non-terminal on the stack and an input symbol, tell us which production to use Empty cells are errors. The rest of the table can be used for descriptive error mesesages Parse error at row, col expecting one of symbols where the predictor table does have entries. What if a cell contains more than one rule The method breaks down. This grammar is called LL 1.L left to right scan of input. L leftmost derivations produced.1 1 symbol of lookahead. top down parsing - starting with s ending at w - recall that triangle diagram. We can automatically compute the predictor table. Predict A, a - rules that apply when A is on the stack and a is the next input character. First , in V is the set of characters thata can be the first letter of a derivation starting from. For example First AyB. So Predict A, a.However, this is not quite right What if A. Then a might not come from A, but from something AFTER A. We missed something - what if there is more than one production with the same LHS How do we know which one to pick. So, Predict A, a U. Nullable true if , false otherwise. Follow A the set of terminal symbols that can come immediately after A in a derivation of S. In our example Follow Aputing Nullable - Nullable true if . Mon Dec 13 2010 Andreas Schwab - 2 12 2-1- Update to 2 12 2 release - Declare wcpcpy and wcpncpy only under GNUSOURCE - Fix use of restrict in wchar h and string h - Fix race in qsortr initialization BZ 11655 - Fix IPTOSCLASS definition BZ 11903 - Don t ignore zero TTL in DNS answers - Fix x86 pthreadcondsignal FUTEXWAKEOP fallback - Work around kernel rejecting valid absolute timestamps - Allow auxcachefile open ing to fail silently even in the chroot mode BZ 11149 - Fix multiple nsscompat initgroups bugs BZ 10085 - Properly convert ffsid in statvfs BZ 11611 - Define MAPHUGETLB and SWAPFLAGDISCARD - Avoid too much stack use in fnmatch BZ 11883 - Fix comparison in sqrtl for IBM long double 128 - Fix warnings in bswap16 BZ 12194 - Properly quote output of locale BZ 11904 - Fix perturbing in malloc on free BZ 12140 - Don t expand DST twice in dlopen - Fix memory leak for some invalid regular expressions BZ 12078 - Linux getifaddrs might return entries with - ifaaddr being NULL BZ 12093 - Handle large malloc requests BZ 12005 - getdents64 fallback dtype support - Fix register conflict in s390 longjmpchk - Fix FORITYSOURCE version of longjmp for Linux x86-64 BZ 11968 - Fix array overflow in floating point parser BZ 7066 - Missing server address again leads to localhost being used BZ 10851 - Document MPERTURB - Fix vDSO synthetic hwcap handling so they are not masked out from matching - 32bit memset-sse2 S fails with uneven cache size BZ 12191 - Verify in ttyname that the symlink is valid BZ 12167 - Fix concurrency problem between dlopen and dliteratephdr - Fix x86-64 strchr propagation of search byte into all bytes of SSE register BZ 12159 - Fix alignment of AVX safe area on x86-64 BZ 12113, 643889 - Don t ignore ORIGIN in libraries Fri Oct 22 2010 Andreas Schwab - 2 12 1-4- Require suid bit on audit objects in privileged programs CVE-2010-3856 Tue Oct 19 2010 Andreas Schwab - 2 12 1-3- Update from 2 12 branch - Fix strstr and memmem algorithm BZ 12092, 641124 - Fix han dling of tail bytes of buffer in SSE2 SSSE3 x86-64 version strncmp BZ 12077 - Never expand ORIGIN in privileged programs 643306, CVE-2010-3847 Fri Aug 20 2010 Andreas Schwab - 2 12 1-2- Update from 2 12 branch - Fix ifunc thunk for strspn on x86 in static libc Tue Aug 17 2010 Andreas Schwab - 2 12 1-1- Update from 2 12 branch - 2 12 1 release - Fix error handling in getloginr 580709 Tue Jul 06 2010 Andreas Schwab - 2 12-3- Update from 2 12 branch - Fix use of extendalloca in NIS - Fix a couple of REDIRECT THROW occurrences - Workaround assembler bug sneaking in nopl 579838 Tue Jun 01 2010 Andreas Schwab - 2 12-2- Update from 2 12 branch - Correct x86 CPU family and model check BZ 11640, 596554 - Don t crash on unresolved weak symbol reference - Implement recvmmsg also as socketcall - sunrpc Fix spurious fall-through - Make compatible with C 593762 - Enable IDN support in getent - Fix race in free sanity check 594784 - Fix lookup of collation sequence value during regexp matching - Fix n ame of locale 589138 - Handle too-small buffers in Linux getloginr BZ 11571, 589946 - Fix users and groups creation in nscd post script - Require coreutils instead of sh-utils - Fix typo causing missing directory ownership Tue May 04 2010 Roland McGrath - 2 12-1- Update to 2 12 release - Fix ldconfig chroot handling - Don t deadlock in dliteratephdr while un loading objects - Fix handling of newline in addmntent - Fix AIO when thread creation failed Fri Apr 16 2010 Andreas Schwab - 2 11 90-20- Update from master - Fix bugs in x86-32 strcmp-sse4 S and strcmp-ssse3 S - Add x86-32 FMA support - Don t crash in trace mode when dependencies are missing - x86-64 SSE4 optimized memcmp - Fix makecontext on s390 s390x Tue Apr 13 2010 Andreas Schwab - 2 11 90-19- Avoid multiarch memcmp in tzdata-update 581677 Mon Apr 12 2010 Andreas Schwab - 2 11 90-18- Update from master - Implement interfaces to set and get names of threads BZ 11390 - Locale data updates BZ 10824, BZ 10936, BZ 11470, BZ 11471 - P rint reload count in nscd statistics BZ 10915 - Fix reading loginuid file in getlogin - Fix fallocate error return on i386 - Fix cproj implmentation BZ 10401 - Fix getopt handing BZ 11039, BZ 11040, BZ 11041 - Implement new mode for NIS table BZ 11134 - Obey LDHWCAPMASK in lookups Tue Apr 06 2010 Andreas Schwab - 2 11 90-17- Update from master - Locale data updates BZ 11007, BZ 11258, BZ 11272, BZ 10554 - Handle DNS timeouts in old-style lookup code BZ 11010 - Fix aux cache handling in ldconfig with chroot BZ 11149 - Fix printing error messages in getopt BZ 11043 - Declare iruserok and iruserokaf BZ 11070 - Fix option aliasing in argp BZ 11254 - Handle POSIX-compliant errno value of unlink in remove BZ 11276 - Fix definition and testing of SISSOCK BZ 11279 - Fix retrieving of kernel header version BZ 11287 - Fix concurrent handling of cpufeatures BZ 11292 - Handle unnecessary padding in getdents64 BZ 11333 - Fix changes to interface list during getifaddrs calls BZ 11387 - Missing memor y barrier in DES initialization BZ 11449 - Fix spurious UNAVAIL status is getaddrinfo - Add support for new clocks BZ 11389 - Fix Linux getlogin implementation - Fix missing zero-termination in cuserid BZ 11397 - Fix glob with empty pattern - Fix handling of STBGNUUNIQUE in LDTRACEPRELINKING - Unify wintt handling in wchar h and wctype h BZ 11410 - Implement handling of libc ABI in ELF header - Don t underestimate length of DST substitution in rpath - Power7-optimized 64-bit and 32-bit memcpy - Assign global scope to RFC 1918 addresses 577626 Thu Mar 18 2010 Andreas Schwab - 2 11 90-16- Fix SSSE3 memcmp 574210 Tue Mar 09 2010 Andreas Schwab - 2 11 90-15- Update from master - sparc64 Fix handling of RSPARCTLSLE relocations 571551 - Handle ext4 and logfs in statvfs functions - Fix setxid race with thread creation - Pass - mtune i686 to assembler when compiling for i686 - Fix RX8664PC32 overflow detection - Fix msgrcv on sparc64 - Fix unwind info in x86 strcmp-sse4 S BZ 11332 - sparc Add mu ltiarch support for memset bzero memcpy - Remove directories owned by filesystem 569414 - Add ghost to glibc-common 567748 Tue Feb 23 2010 Andreas Schwab - 2 11 90-14- Update from master - Sparc updates - Fix SSSE3 memcpy 556584 Mon Feb 22 2010 Andreas Schwab - 2 11 90-13- Update from master - Use CPUIDOFFSET instead of FEATUREOFFSET - Add 32bit memcmp strcmp strncmp optimized for SSSE3 SSS4 2 - Fix file descriotor leak in nftw with FTWCHDIR BZ 11271 - Add Sparc STTGNUIFUNC support - Add power7-optimized classification functions - Reapply Optimize 32bit memset memcpy with SSE2 SSSE3 - Use unsigned comparison in sse memcpy memset Mon Feb 08 2010 Andreas Schwab - 2 11 90-12- Update from master - Update constants in for current kernels 11235 - Fix endless loop with invalid etc shells file 11242 - Fix sorting of malayalam letter na 10414 - Add kokIN locale - Use common collation data in asIN locale - Avoid alloca in setenv for long strings - Use shared mapping to reserve memory when creating l ocale archive 10855 - Fix fstat on Linux sparc64 11155 Mon Feb 01 2010 Andreas Schwab - 2 11 90-11- Update from master - Fix error checking in iconv 558053 - Don t map U00DF to U1E9E in toupper table - nlloadlocale incorrectly handles mmap failures BZ 11200 - Fix various issues in regex matcher BZ 11183, BZ 11184, BZ 11185, BZ 11186, BZ 11187, BZ 11188, BZ 11189, BZ 11190, BZ 11191, BZ 11192, BZ 11193 Tue Jan 19 2010 Andreas Schwab - 2 11 90-10- Update from master - Fix longjmpchk for s390 s390x - Remove duplicate definitions of ODSYNC and ORSYNC for Linux sparc - Ignore negative dynamic entry types 546890 - Fix pthreadcond wait with requeue-PI on i386 548989 - Fix XOPENSOURCEEXTENDED handling - Revert Optimize 32bit memset memcpy with SSE2 SSSE3 Fri Jan 15 2010 Andreas Schwab - 2 11 90-9- Update from master - Define IPTOSCLASS macros according to RFC 2474 BZ 11027 - Always use IPv4 sockets for IPv4 addresses BZ 11141 - regcomp c do not ignore memory allocation failure BZ 11127 - Fix ma llocinfo without prior allocations BZ 11126 - Optimize 32bit memset memcpy with SSE2 SSSE3 - Relax feature tests in headers Tue Jan 12 2010 Andreas Schwab - 2 11 90-8- Update from master - More POSIX conformance fixes Mon Jan 11 2010 Andreas Schwab - 2 11 90-7- Fix build failure Mon Jan 11 2010 Andreas Schwab - 2 11 90-6- Update from master - POSIX conformance fixes BZ 11125 Mon Jan 04 2010 Andreas Schwab - 2 11 90-5- Update from master - Additional setcontext , etc conformance tests BZ 11115 - Handle ATFDCWD in futimens BZ 10992 - Update poll h header for POSIX 2008 BZ 11093 - Avoid ELF lookup race Mon Dec 14 2009 Andreas Schwab - 2 11 90-4- Update from master - Add Requeue-PI support for x86 arch - Redefine OSYNC and ODSYNC to match 2 6 33 kernels - Fix a few error cases in name4r lookup handling BZ 11000 - Fix kernel version check in recent ptsname change BZ 11046 - Add more warnings to exec functions BZ 11056 - Add recvmmsg interface - Define SCHEDIDLE and SCHEDRESETONFORK for Linu x Mon Nov 30 2009 Andreas Schwab - 2 11 90-3- Update from master - Fix infloop in pthreaddisableasynccancel on x8664 537690 - Prevent unintended file desriptor leak in grantpt 530558 - Fix startup to security-relevant statically linked binaries 528631 - Re-install CFI in x86 x8664 clone 491542 Tue Nov 24 2009 Andreas Schwab - 2 11 90-2- Update from master - Define week, firstweekday, and firstworkday for enDK locale 525126 - Use struct timespec for timestamps in struct stat also if USEXOPEN2K8 539870 - Fix week information for nlNL locale 499748 - Update ntpgettime for Linux 479558 - Fix getwc and putwc on non-wide streams BZ 10958 - Avoid warnings in CPU macros when using const bitsets BZ 10918 - Handle LCGLOBALLOCALE in duplocale BZ 10969 - Fix NCLOCALENAME definition BZ 10968 - Add missing Linux MADV definitions BZ 10972 - Add support for new Linux error ERFKILL BZ 10939 - Enable multi-arch support on ppc and ppc64 Thu Nov 12 2009 Andreas Schwab - 2 11 90-1- Update from master Thu N ov 05 2009 Andreas Schwab - 2 11-2- Fix readahead on powerpc32 - Fix RPPC64 handling - Fix preadv, pwritev and fallocate for - DFILEOFFSETBITS 64 533063 Mon Nov 02 2009 Andreas Schwab - 2 11-1- Update to 2 11 release - Disable multi-arch support on PowerPC again since binutils is too old - Fix crash in tzdata-update due to use of multi-arch symbol 532128 Fri Oct 30 2009 Andreas Schwab - 2 10 90-27- Update from master - Fix races in setXid implementation BZ 3270 - Implement IFUNC for PPC and enable multi-arch support - Implement mkstemps mkstemps64 and mkostemps mkostemps64 BZ 10349 - Fix IA-64 and S390 sigevent definitions BZ 10446 - Fix memory leak in NIS grp database handling BZ 10713 - Print timestamp in nscd debug messages BZ 10742 - Fix mixing IPv4 and IPv6 name server in - Fix range checks in coshl - Implement SSE4 2 optimized strchr and strrchr - Handle IFUNC symbols in dlsym 529965 - Misc fixes BZ 10312, BZ 10315, BZ 10319, BZ 10391, BZ 10425, BZ 10540, BZ 10553, BZ 10564, BZ 10 609, BZ 10692, BZ 10780, BZ 10717, BZ 10784, BZ 10789, BZ 10847- No longer build with - fno-var-tracking-assignments Mon Oct 19 2009 Andreas Schwab - 2 10 90-26- Update from master - Add longjmpchk for sparc - Avoid installing the same libraries twice Mon Oct 12 2009 Andreas Schwab - 2 10 90-25- Update from master - Fix descriptor leak when calling dlopen with RTLDNOLOAD 527409 - Fix week-1stday in C locale - Check for integer overflows in formatting functions - Fix locale program error handling 525363 Mon Sep 28 2009 Andreas Schwab - 2 10 90-24- Update from master - Fix missing reloc dependency 517001 Mon Sep 21 2009 Andreas Schwab - 2 10 90-23- Update from master Mon Sep 14 2009 Andreas Schwab - 2 10 90-22- Update from master - Fix endless loop in localedef - Fix longjmpchk on s390 s390x - Fix exit codes in nscd start script 521848 - Build with - fno-var-tracking-assignments for now 523172 Mon Sep 07 2009 Andreas Schwab - 2 10 90-21- Update from master - Fix strstr strcasestr on i386 5 19226 Thu Sep 03 2009 Andreas Schwab - 2 10 90-20- Update from master - Fix strstr strcasestr fma fmaf on x8664 519226 - Fix lookup of group names in hesiod initgroups 520472 Wed Sep 02 2009 Andreas Schwab - 2 10 90-19- Update from master - Fix x8664 bits mathinline h for - m32 compilation Tue Sep 01 2009 Andreas Schwab - 2 10 90-18- Update from master - fix parse error in 520209 Thu Aug 27 2009 Roland McGrath - 2 10 90-17- Update from master Wed Aug 26 2009 Andreas Schwab - 2 10 90-16- Update from master - handle AVX saving on x86-64 in interrupted symbol lookups 519081 Mon Aug 24 2009 Andreas Schwab - 2 10 90-15- Update from master - fix fortify failure with longjmp from alternate stack 512103 - Add conflict with prelink 509655 Mon Aug 17 2009 Andreas Schwab - 2 10 90-14- Update from master - fix pthreadcondsignal 516469 Mon Aug 10 2009 Andreas Schwab - 2 10 90-13- Update from master - fix rehashing of unique symbols 515677 - Fix spurious messages with --excludedocs 515948 Mon Aug 03 2009 Andreas Schwab - 2 10 90-12- Update from master - fix fortify failure with longjmp from alternate stack 512103 Thu Jul 30 2009 Andreas Schwab - 2 10 90-11- Update from master - Don t package debuginfo files in glibc-devel Tue Jul 28 2009 Andreas Schwab - 2 10 90-10- Update from master fix memory ordering in pthreadmutexunlock BZ 10418 implement RESUSEDNSSEC option in resolver 205842 fix hang in ldd - r 513945 Mon Jul 27 2009 Andreas Schwab - 2 10 90-9- Update from master Fri Jul 24 2009 Fedora Release Engineering - 2 10 90-8 1- Rebuilt for Fri Jul 24 2009 Jakub Jelinek - 2 10 90-7 1- Fix up pthreadcondtimedwait on x8664 with old kernels Thu Jul 23 2009 Andreas Schwab - 2 10 90-7- Update from master - Build with - DNDEBUG unless using a prerelease Thu Jul 23 2009 Andreas Schwab - 2 10 90-6- Rebuilt with to fix static binaries Wed Jul 22 2009 Andreas Schwab - 2 10 90-5- Update from master - Undefine i686 on x86 to fix build Mon Jul 20 2009 Andreas Schwab - 2 10 90-4- Update from maste r - Don t build separate i686 package Wed Jul 08 2009 Andreas Schwab 2 10 90-3- Reenable setuid on ptchown Thu Jul 02 2009 Andreas Schwab 2 10 90-2- Update from master Fri Jun 26 2009 Andreas Schwab 2 10 90-1- Update from master - Enable multi-arch support on x86 x86-64 - Add requires glibc-headers to glibc-devel 476295 - Implement second fallback mode for DNS requests 505105 - Don t generate invalid POSIX TZ string for Asia Dhaka timezone 506941 - Allow backtrace through longjmpchk on powerpc Fri May 22 2009 Jakub Jelinek 2 10 1-2- fix accept4 on architectures other than i 86 x8664- robustify nscd client code during server GC - fix up nscd segfaults during daemon shutdown - fix memchr on ia64 BZ 10162 - replace the Sun RPC license with the BSD license, with the explicit permission of Sun Microsystems - fix up powerpc long double errno reporting Sun May 10 2009 Jakub Jelinek 2 10 1-1- fix up getsgentr and getsgnamr exports on i 86 and ppc Sat May 09 2009 Jakub Jelinek 2 10-2- update from trunk - glibc 2 10 release - fix memchr on x8664 499689 Mon Apr 27 2009 Jakub Jelinek 2 9 90-22- update from trunk - further localedef fixes - fix build-locale-archive Fri Apr 24 2009 Jakub Jelinek 2 9 90-21- update from trunk - fix localedef - fix SHIFTJIS iconv EILSEQ handling 497267 - misc fixes BZ 10093, BZ 10100 Fri Apr 24 2009 Jakub Jelinek 2 9 90-20- update from trunk - fix p v 497429, 497434 - fix strfmon 496386 Thu Apr 16 2009 Jakub Jelinek 2 9 90-19- update from trunk - fix dlopen from statically linked binaries 495830 Thu Apr 16 2009 Jakub Jelinek 2 9 90-18- update from trunk - fix fallocate Wed Apr 15 2009 Jakub Jelinek 2 9 90-17- update from trunk - if threads have very small stack sizes, use much smaller buffer in getnprocs when called from within malloc 494631 Tue Apr 14 2009 Jakub Jelinek 2 9 90-16- update from trunk Thu Apr 09 2009 Jakub Jelinek 2 9 90-15- rebuilt with fixed gcc to avoid miscompilation of i586 memmove - reenable experimental malloc again Wed Apr 08 2009 Jakub Jelinek 2 9 90-14- update from trunk - temporarily disable experimental malloc Tue Apr 07 2009 Jakub Jelinek 2 9 90-13- update from trunk - fix strverscmp 494457 - configure with --enable-nss-crypt Wed Apr 01 2009 Jakub Jelinek 2 9 90-12- update from trunk - configure with --enable-experimental-malloc Fri Mar 20 2009 Jakub Jelinek 2 9 90-11- update from trunk - POSIX 2008 prototype adjustments for scandir , alphasort and versionsort - fix libthreaddb 491197 Tue Mar 10 2009 Jakub Jelinek 2 9 90-10- update from trunk - fix atexit cxaatexit Mon Mar 09 2009 Jakub Jelinek 2 9 90-9- update from trunk - POSIX 2008 support - DXOPENSOURCE 700 and - DPOSIXCSOURCE 200809L - move libnldblnonshared a on ppc s390 sparc back to glibc-devel Fri Feb 27 2009 Roland McGrath - 2 9 90-8 1- fix libthreaddb 487212 Tue Feb 24 2009 Fedora Release Engineering - 2 9 90-8- Rebuilt for Wed Feb 18 2009 Jakub Jelinek 2 9 90-7- update from trunk - adjust for i586 i686 from i386 i686 build - split static libraries into glibc-static subpackage - ld - r the whole libpthread a together to avoid endless issues with - static - lpthread - require 2 6 18 and later kernel Wed Feb 04 2009 Jakub Jelinek 2 9 90-3- update from trunk - ISO C compliant strchr etc with GCC 4 4 - ATRANDOM support Thu Jan 08 2009 Jakub Jelinek 2 9 90-2- update from trunk Fri Jan 02 2009 Jakub Jelinek 2 9 90-1- update from trunk 478314 Mon Dec 08 2008 Jakub Jelinek 2 9-3- temporarily disable nssdnsgethostbyname4r 459756 - NIS hostname lookup fixes 473073, 474800, BZ 7058 - fix unsetenv 472941 Thu Nov 13 2008 Jakub Jelinek 2 9-2- glibc 2 9 release - fix CPUALLOCSIZE on 32-bit arches BZ 7029 Wed Nov 12 2008 Jakub Jelinek 2 8 90-17- update from trunk - don t abort on broken DNS replies 469299, BZ 7009 - misc fixes BZ 6966, BZ 7008, BZ 6955, BZ 6843 Fri Oct 31 2008 Jakub Jelinek 2 8 90-16- update from trunk - further resolver fixes - another dynamic TLS handling fix 469263 - misc fixes BZ 6867, BZ 6875, BZ 6919, BZ 6920, BZ 6942, BZ 6947, BZ 6 968, BZ 6974, BZ 6980, BZ 6995 - rebuild with newer rpm to avoid stripping shared libraries when they shouldn t be 468129 Tue Oct 28 2008 Jakub Jelinek 2 8 90-15- update from trunk - libcresnquery fixes 466786 Sun Oct 19 2008 Jakub Jelinek 2 8 90-14- update from trunk - fix dynamic TLS handling 467309 - fix sys signalfd h for C 467172 - fix sprof 458861 - fix mcount and socket syscalls on s390x 464146 - try harder to allocate memory in valloc and pvalloc 461481 - fix power6 32-bit libs 467311 Fri Oct 10 2008 Dennis Gilmore 2 8 90-13- apply sparcv9v memset patch from jakub and davem Fri Aug 29 2008 Jakub Jelinek 2 8 90-12- update from trunk - revert origin changes 457849 - use MAPSTACK for thread stacks - misc fixes BZ 6845, BZ 6544, BZ 6634, BZ 6589, BZ 6790, BZ 6791, BZ 6824 - power7 bits BZ 6817 - fix expm1 on i 86 x8664 43354, BZ 5794 Sat Aug 02 2008 Jakub Jelinek 2 8 90-11- update from trunk - fix non-absolute ORIGIN handling 457560 - exported some further libresolv APIs 453325 - m isc fixes Tue Jul 29 2008 Jakub Jelinek 2 8 90-10- update from trunk - resolver fixes - misc fixes BZ 6771, BZ 6763, BZ 6698, BZ 6712 - s390 utmp utmpx bi-arch support BZ 6724 - popen e flag - frFR locale changes reenabled Wed Jul 16 2008 Jakub Jelinek 2 8 90-9- update from trunk - fix unbuffered vfprintf if writing to the stream fails 455360 - remove useless malloc using debugging hooks message 455355 - nscd fixes - fix resolver alignment issues 454500 - fix setvbuf BZ 6719 Thu Jul 03 2008 Jakub Jelinek 2 8 90-8- update from trunk - watch even in nscd using inotify - some nscd fixes Fri Jun 13 2008 Jakub Jelinek 2 8 90-7- update from trunk - avoid lround on ppc clobbering cr3 cr4 registers 450790 - further nscd fixes 450704 - use inotify in nscd to watch files Thu Jun 12 2008 Jakub Jelinek 2 8 90-6- update from trunk - nscd fixes 450704 - fix getservbyport 449358 - fix regexp h 446406 - avoid crashing on TDNAME in DNS responses 450766 Sun May 25 2008 Jakub Jelinek 2 8 90-5- update from trunk Tue May 20 2008 Jakub Jelinek 2 8 90-4- further getaddrinfo and nscd fixes Sun May 18 2008 Jakub Jelinek 2 8 90-3- getaddrinfo and nscd fixes - reenable assertion checking in rawhide Fri May 16 2008 Jakub Jelinek 2 8 90-2- fix getaddrinfo 446801, 446808 Thu May 15 2008 Jakub Jelinek 2 8 90-1- update to trunk - O n memmem strstr strcasestr - i386 x8664 TLS descriptors support - concurrent IPv4 and IPv6 DNS lookups by getaddrinfo Mon May 05 2008 Jakub Jelinek 2 8-3- don t run telinit u in post if both dev initctl and sbin initctl exist 444978 - workaround GCC ppc64 miscompilation of c, acosh, atan l 444996 Wed Apr 30 2008 Jakub Jelinek 2 8-2- fix nscd races during GC BZ 5381 - rebuilt with fixed GCC to fix regex miscompilation on power6- SPARC fixes Sat Apr 12 2008 Jakub Jelinek 2 8-1- 2 8 release Fri Apr 11 2008 Jakub Jelinek 2 7 90-16- update to trunk - misc fixes BZ 4997, BZ 5741 - make sure all users of libcsetlocalelock know it is now a rwlock - fix ppc ppc64 compatibility syse rrlist and syssiglist symbols Thu Apr 10 2008 Jakub Jelinek 2 7 90-15- update to trunk - misc fixes BZ 4314, BZ 4407, BZ 5209, BZ 5436, BZ 5768, BZ 5998, BZ 6024 - restart sshd in post when upstart is used - it doesn t have dev initctl 441763 - disable assert checking again Tue Apr 08 2008 Jakub Jelinek 2 7 90-14- update to trunk - misc fixes BZ 5443, BZ 5475, BZ 5478, BZ 5939, BZ 5979, BZ 5995, BZ 6004, BZ 6007, BZ 6020, BZ 6021, BZ 6042 - change mtrace to keep perl 5 10 quiet 441082 - don t share conversion state between mbtowc and wctomb 438687 - if stblksize is too large and malloc fails, retry with smaller buffer in opendir 430768 - correct printf overflow test 358111 Fri Mar 28 2008 Jakub Jelinek 2 7 90-13- update to trunk - don t define ARGMAX in., as it is no longer constant - use sysconf SCARGMAX to get the current argument size limit - fix build on sparc64- only service sshd condrestart if etc rc d init d sshd exists 428859 Wed Mar 26 2008 Jakub Jelinek 2 7 90-12- update to t runk - new CLONE flags in 438542 - nis errno clobbering fix 437945 - fix adjtime 437974 Fri Mar 14 2008 Jakub Jelinek 2 7 90-11- update to trunk - remove define XOPENSTREAMS -1 436349 Wed Mar 05 2008 Jakub Jelinek 2 7 90-8- update to trunk - printf and obstack printf fortification 435905 - fix getnameinfo gethostbyaddr 428067, BZ 5790 - fix yporder 435519, BZ 5854 - misc fixes BZ 5779, BZ 5736, BZ 5627, BZ 5818, BZ 5012 - merge review cleanup Tom Callaway, 225806 Sat Feb 16 2008 Jakub Jelinek 2 7 90-7- update to trunk - make NIMAXHOST and NIMAXSERV available even in BSDish namespaces BZ 5737 - timerfd syscalls Fri Feb 01 2008 Jakub Jelinek 2 7 90-6- fix build Thu Jan 31 2008 Jakub Jelinek 2 7 90-5- update to trunk - rebuild with gcc 4 3 Fri Jan 11 2008 Jakub Jelinek 2 7 90-4- update to trunk - misc fixes BZ 5541, BZ 5545, BZ 5553, BZ 5112, BZ 5520 - getaddrinfo fixes - signalize EOVERFLOW from sempost instead of overflowing the counter - fix i 86 makecontext - fix iconv for iso-2022-jp t ranslit 397021 Thu Jan 03 2008 Jakub Jelinek 2 7 90-3- update to trunk - fix recognition of interface family 425768 - add THROW to ctype loc prototypes Thu Dec 27 2007 Jakub Jelinek 2 7 90-2- update to trunk - nsswitch fix 425768 - temporarily enable assert checking Wed Dec 12 2007 Jakub Jelinek 2 7 90-1- update to trunk - fix USESTRINGINLINES on i 86 408731, 371711 - fix scanf 388751 Wed Oct 17 2007 Jakub Jelinek 2 7-1- glibc 2 7 release - fix tzfile c for times after last transition 333561 - fix sempostAATTGLIBC2 0 on i 86- appease valgrind in initialization - misc fixes BZ 3425, BZ 5184, BZ 5186 Mon Oct 15 2007 Jakub Jelinek 2 6 90-21- fix getgr with nscd Sun Oct 14 2007 Jakub Jelinek 2 6 90-20- install 330031 - disable - DFORTIFYSOURCE support with a warning for GCC 3 4 x and earlier 327641 - plPL locale changes BZ 4098, 242296 - misc fixes BZ 1140, BZ 3195, BZ 3242, BZ 4359 Thu Oct 11 2007 Jakub Jelinek 2 6 90-19- fix - simple preprocessor in localedef, fix deDE collation with it Wed Oct 10 2007 Jakub Jelinek 2 6 90-18- add signalfd, eventfd, eventfdread, eventfdwrite - qsort speedups - workaround for cpuid bugs 324081 - make sure gettext s conversionlock is initialized even if program isn t linked against only dlopens it 321761 - misc fixes BZ 5112, BZ 5113, BZ 5104, BZ 5063, BZ 5010, BZ 4407, BZ 3924, BZ 5103, BZ 2633, BZ 181, BZ 73, 321901 Wed Oct 03 2007 Jakub Jelinek 2 6 90-17- fix swprintf with - DFORTIFYSOURCE 1 - mlong-double-64 on ppc s390 sparc - strcoll fixes - misc fixes BZ 645, BZ 5071 - locale fixes BZ 4941, 299321, 203364, 196711, 236212 Sat Sep 29 2007 Jakub Jelinek 2 6 90-16- misc fixes BZ 4963, BZ 4972, BZ 5028, BZ 5043, BZ 5058 - improve - DFORTIFYSOURCE diagnostic through warning error attributes - fix wcscpy, wcpcpy, fgetws, fgetwsunlocked, swprintf and vswprintf fortification inlines - fix a scalability issue with lazy binding in heavily multithreaded programs Thu Sep 20 2007 Jakub Jelinek 2 6 90-15- 5 SHA-256 and 6 SHA-512 support in crypt 228697, 2 49477, 173834 Tue Sep 18 2007 Jakub Jelinek 2 6 90-14- - DFORTIFYSOURCE support for C - fortification of fread - support scanf m allocation modifier ms, mls, mc - in - std c99 or - DXOPENSOURCE 600 mode don t recognize as, aS and a as a GNU extension for scanf - fix splice, vmsplice, tee return value, make them cancellation points - mqopen checking - use inline function rather than function-like macro for open checking - IFAFOPTIMISTIC handling in getaddrinfo 259681 - fix an ABBA deadlock in 284171 - remove sparc unwind info from start and clone Mon Aug 27 2007 Jakub Jelinek 2 6 90-13- fix personality on x8664 ppc ppc64 256281 Sat Aug 25 2007 Jakub Jelinek 2 6 90-12- readd x8664 gettimeofday stuff, initialize it earlier - nislist fix 254115 - workaround for bugs in ia64 silly emul ia32-linux hack 253961 - misc fixes BZ 3924, BZ 4566, BZ 4582, BZ 4588, BZ 4726, BZ 4946, BZ 4905, BZ 4814, BZ 4925, BZ 4936, BZ 4896, BZ 4937, BZ 3842, BZ 4554, BZ 4557, BZ 4938 Fri Aug 17 2007 Jakub Jelinek 2 6 90- 11- remove strtoldinternal and wcstoldinternal from ppc s390 sparc - ldbl h headers - temporarily backout x8664 gettimeofday S changes 252453 - some further sparc, sparc64 and alpha fixes Wed Aug 15 2007 Jakub Jelinek 2 6 90-10- don t open with ONOATIME 252146 - s390 , alpha and sparc fixes - sparcv9 is no longer an aux arch, as we expect to not build glibc any longer, only and new two aux arches and Tue Aug 14 2007 Jakub Jelinek 2 6 90-9- private futex even for mutexes and condvars - some further OCLOEXEC changes - use vDSO on x8664 if available - ia64 build fixes 251983 Fri Aug 10 2007 Roland McGrath 2 6 90-8- update to trunk - fix missing strtoldl export on ppc64 Thu Aug 09 2007 Roland McGrath 2 6 90-6- update to trunk - fix local PLT regressions - spec file revamp for new Sun Aug 05 2007 Jakub Jelinek 2 6 90-4- fix and on ppc32, so that it is not using bss PLT Sat Aug 04 2007 Jakub Jelinek 2 6 90-3- fix open macro for - pedantic 250897 - add transliteration for l with stroke 250492 - fix s trtod -0 , NULL - update License tag Wed Aug 01 2007 Jakub Jelinek 2 6 90-2- make aux-cache purely optional performance optimization in ldconfig, don t issue any errors if it can t be created 250430 - remove overrideheaders hack, BuildRequire 2 6 22 kernel-headers and rely on its content Tue Jul 31 2007 Jakub Jelinek 2 6 90-1- update to trunk - private futex optimizations - open argument checking - ldconfig speedups Sun Jul 08 2007 Jakub Jelinek 2 6-4- filter pseudo-files from debuginfo source lists 245714 - fix sscanf when errno is EINTR before the call BZ 4745 - save restore errno around reading etc default nss BZ 4702 - fix LDHWCAPMASK handling - disable workaround for 210748, instead backport locking fixes from the trunk 235026 - new x8664 memcpy - don t write uninitialized padding bytes to nscd socket - fix dl sym, dliteratephdr and dlopen if some library is mapped into inter-segment hole on x8664 245035, 244545 - fix LDAUDIT a b program 180432 - don t crash on pseudo-zero long double values passed to printf on i 86 x8664 ia64 BZ 4586 - fix printf La and strtold with some hexadecimal floating point constants on ppc ppc64- fix nextafterl on ppc ppc64- fix semtimedwait on i 86 and x8664 Thu May 24 2007 Jakub Jelinek 2 6-3- don t use config missingok for instead of removing it altogether truncate it to zero size 240697 - add a workaround for 210748 Mon May 21 2007 Jakub Jelinek 2 6-2- restore mallocsetstate backwards compatibility 239344 - fix epollpwait BZ 4525 - fix printf with unknown format spec or positional arguments and large width and or precision BZ 4514 - robust mutexes fix BZ 4512 Tue May 15 2007 Roland McGrath 2 6-1- glibc 2 6 release Fri May 11 2007 Jakub Jelinek 2 5 90-24- utimensat, futimens and lutimes support Thu May 10 2007 Jakub Jelinek 2 5 90-23- use madvise MADVDONTNEED in malloc - fix ia64 feraiseexcept - fix s390 feholdexcept BZ 3427 - ppc fenv fixes - make fdatasync a cancellation point BZ 4465 - fix printf for huge precisions with wide char code and multi-byte strings - fix dladdr 232224, BZ 4131 Fri May 04 2007 Jakub Jelinek 2 5 90-22- add transliteration for BZ 3213 - fix scanf with f on hexadecimal floats without exponent BZ 4342 - fix printf with very large precisions for s 238406, BZ 4438 - fix inetntop size checking for AFINET BZ 4439 - for printf e avoid 1 000e-00, for exponent 0 always use sign 238431 - fix a regression introduced in 223467 changes - gethostby r alignment fixes BZ 4381 - fix ifaddrs error handling Mon Apr 16 2007 Jakub Jelinek 2 5 90-21- don t include individual locale files in glibc-common, rather include prepared locale-archive template and let build-locale-archive create locale-archive from the template and any user supplied usr lib locale directories, then unlink the locale-archive template - this should save 80MB of glibc-common occupied disk space - fix XOPENVERSION BZ 4364 - fix printf with g and values tiny bit smaller than 1 e-4 235864, BZ 4362 - fix NIS nisfindserver 235229 Sat Mar 31 2007 Jaku b Jelinek 2 5 90-20- assorted NIS speedups 223467 - fix HAVELIBCAP configure detection 178934 - remove sbin rpcinfo from glibc-common 228894 - nexttoward nextafter fixes BZ 3306 - feholdexcept feupdateenv fixes BZ 3427 - speed up fnmatch with two or more in the pattern Sat Mar 17 2007 Jakub Jelinek 2 5 90-19- fix power6 libm compat symbols on ppc32 232633 - fix child refcntr in NPTL fork 230198 - fix ifaddrs with many net devices on 4KB page size arches 230151 - fix pthreadmutextimedlock on x8664 228103 - various fixes BZ 3919, BZ 4101, BZ 4130, BZ 4181, BZ 4069, BZ 3458 Wed Feb 21 2007 Jakub Jelinek 2 5 90-18- fix nftw with FTWCHDIR on BZ 4076 - nscd fixes BZ 4074 - fix fmod on i 86 BZ 3325 - support localized digits for fp values in scanf BZ 2211 - namespaces fixes BZ 2633 - fix euidaccess BZ 3842 - glob fixes BZ 3996 - assorted locale data fixes BZ 1430, BZ 672, BZ 58, BZ 3156, BZ 2692, BZ 2648, BZ 3363, BZ 3334, BZ 3326, BZ 3322, BZ 3995, BZ 3885, BZ 3884, BZ 3851 Sun Feb 11 2007 J akub Jelinek 2 5 90-17- RFC2671 support in resolver 205842 - fix strptime BZ 3944 - fix regcomp with REGNEWLINE BZ 3957 - fix pthreadmutextimedlock on x8664 228103 Fri Feb 02 2007 Jakub Jelinek 2 5 90-16- add strerrorl - fix application crashes when doing NSS lookups through nscd mmapped databases and nscd decides to start garbage collection during the lookups 219145, 225315 - fix 0lld printing of 0LL on 32-bit architectures BZ 3902 - ignore errors from install-info in glibc-devel scriptlets 223691 Wed Jan 17 2007 Jakub Jelinek 2 5 90-15- fix NIS getservbyname when proto is NULL - fix nsscompat group handling 220658 - cache services in nscd - fix double free in ftsclose 222089 - fix vfork execvp memory leak 221187 - soft-fp fixes BZ 2749 - further strtod fixes BZ 3855 - make sure pthreadkill doesn t return EINVAL even if the target thread exits in between pthreadkill ESRCH check and the actual tgkill syscall 220420 - fix ABBA deadlock possibility in scope locking code Tue Dec 19 2006 Jaku b Jelinek 2 5 90-14- fix rand48 on 64-bit arches BZ 3747 - handle power6x ATPLATFORM 216970 - fix a race condition in getXXbyYYr 219145 - fix tst-pselect testcase Thu Dec 14 2006 Jakub Jelinek 2 5 90-13- fix setcontext on ppc32 219107 - fix wide stdio after setvbuf 217064, BZ 2337 - handle relatime mount option in statvfs - revert i 86 x8664 clone CFI temporarily Sun Dec 10 2006 Jakub Jelinek 2 5 90-12- fix hasmntopt 218802 - fix setusershell and getusershell 218782 - strtod fixes BZ 3664, BZ 3673, BZ 3674 - fix memusage with realloc x, 0 Tue Dec 05 2006 Jakub Jelinek 2 5 90-11- allow suid apps to setenv NISPATH and influence through that nislist and nislookup 209155 - fix ttyname and ttynamer with invalid file descriptor 218276 - csCZ LCTIME fixes 218438 - fix build with 2 6 19 headers 217723 Fri Dec 01 2006 Jakub Jelinek 2 5 90-10- fix x86-64 restorert unwind info Thu Nov 30 2006 Jakub Jelinek 2 5 90-9- fix last svcrun change 217850 - on ppc64 build libcstartmain without unwind info, as it breaks MDFROBUPDATECONTEXT 217729, 217775 in the future that could be fixable just by providing r2 in libcstartmain instead - add unwind info for x86-64 restorert signal return landing pad 217087 - add power6x subdir to and rtkaio , link all libs from power6 into them Tue Nov 28 2006 Jakub Jelinek 2 5 90-8- fix svcrun 216834, BZ 3559 - add - fasynchronous-unwind-tables to CFLAGS 216518 - make sure there is consistent timestamp for etc localtime and etc rpc between multilib glibc rpms Mon Nov 20 2006 Jakub Jelinek 2 5 90-7- handle IPv6 addresses in etc hosts that are mappable to IPv4 addresses in IPv4 host lookups 215283 - fix include etc alias handling 215572 - handle new tzdata format to cope with year 2037 transitions on 64-bit architectures Fri Nov 10 2006 Jakub Jelinek 2 5 90-6- fix strxfrm fix - fix i 86 floor and ceil inlines BZ 3451 Thu Nov 09 2006 Jakub Jelinek 2 5 90-5- fix sysconf SCLEVEL CACHESIZE on Intel Core Duo CPUs - fix on TLSDTVATTP architectures - fix --inhibit-rpa th 214569 - fix rdebug content when prelinked executes a program as its argument - fix strxfrm - powerpc-cpu add-on updates Fri Nov 03 2006 Jakub Jelinek 2 5 90-4- fix atexit backwards compatibility 213388 - add maiIN locale 213415 - remove bogus symlink 213555 - fix memusage 213656 - change category 209493 Sun Oct 29 2006 Jakub Jelinek 2 5 90-3- fix suid sgid binaries on i 86 x8664 212723 Fri Oct 27 2006 Jakub Jelinek 2 5 90-2- fix ia64 build - don t call dlclose outside of dlloadlock critical section if dlopen failed BZ 3426 - add rtld scope locking 211133 Wed Oct 25 2006 Jakub Jelinek 2 5 90-1- fix i 86 6 argument syscalls e g splice - fix rtld minimal realloc BZ 3352 - fix RFC3484 getaddrinfo sorting according to rules 4 and 7 BZ 3369 - fix xdrmemsetpos 211452 - bump GLIBCMINOR - increase PTHREADSTACKMIN on ppc to 128K to allow 64K pagesize kernels 209877 - speed up initgroups on NIS 208203 Mon Oct 02 2006 Jakub Jelinek 2 5-2- fix nscd database growing 207928 - bypass prelinking when L DDYNAMICWEAK 1 is in the environment Fri Sep 29 2006 Jakub Jelinek 2 5-1- glibc 2 5 release Wed Sep 27 2006 Jakub Jelinek 2 4 90-36- rebuilt with gcc-4 1 1-26 to fix unwind info Mon Sep 25 2006 Jakub Jelinek 2 4 90-35- fix glob with large number of matches BZ 3253 - fix fchownat on kernels that don t support that syscall BZ 3252 - fix lrintl on s390 Sat Sep 23 2006 Jakub Jelinek 2 4 90-34- fix ppc longjmp BZ 3225 - fix user visible spelling errors BZ 3137 - fix l rint around zero BZ 2592 - avoid stack trampoline in s390 makecontext Fri Sep 15 2006 Jakub Jelinek 2 4 90-33- fix dlclose 206639 - don t load platform optimized libraries if kernel doesn t set ATPLATFORM - fix ppc use - mtune generic even for glibc-devel i386 206437 - fix symlink Fri Sep 15 2006 Jakub Jelinek 2 4 90-32- on ppc use just ATPLATFORM and altivec ATHWCAP bit for library selection - fix lrintl and lroundl on ppc - use hidden visibility on fstatat and mknodat in libcnonshared a Sun Sep 10 2006 Jakub Jelinek 2 4 90-31- fix pthreadcond wait cancellation BZ 3123 - fix lrint on ppc32 BZ 3155 - fix malloc allocating more than half of address space BZ 2775 - fix mktime on 32-bit arches a few years after 2038 BZ 2821 Thu Sep 07 2006 Jakub Jelinek 2 4 90-30- add librtkaio, to use it add rtkaio to your LDLIBRARYPATH or fix orIN February name 204730 - fix pthreadcreate called from cancellation handlers BZ 3124 - fix regex case insensitive searches with characters where upper and lower case multibyte representations have different length e g I and dotless i, 202991 Tue Sep 05 2006 Jakub Jelinek 2 4 90-29- randomize resolver query ids before use instead after use 205113 - fix resolver symver checking with DTGNUHASH 204909 - put section in glibc libraries at the end of RO segment when is present Thu Aug 31 2006 Jakub Jelinek 2 4 90-28- another malloc doubly linked list corruption problem fix 204653 Thu Aug 31 2006 Jakub Jelinek 2 4 90-27- allow LIB and PLATFORM in dlopen parameters even in suid sgid 204399 - han dle LIB PLATFORM in LDLIBRARYPATH - fix splice prototype 204530 Mon Aug 28 2006 Jakub Jelinek 2 4 90-26- real fix for the doubly linked list corruption problem - try harder in realloc to allocate memory BZ 2684 - fix getnameinfo error reporting 204122 - make localedef more robust on invalid input 203728 Fri Aug 25 2006 Jakub Jelinek 2 4 90-25- temporarily back out code to limit number of unsorted block sort iterations 203735, 204027 - handle PLT symbols in dladdr properly BZ 2683 - avoid malloc infinite looping for allocations larger than the system can allocate 203915 Tue Aug 22 2006 Jakub Jelinek 2 4 90-23- malloc fixes, especially for 32-bit arches 202309 - further IN locale fixes 200230 - fix get ent if NIS map is empty 203237 - fix usr bin iconv 203400 Fri Aug 18 2006 Jakub Jelinek 2 4 90-22- rebuilt with latest binutils to pick up 64K - z commonpagesize on ppc ppc64 203001 Tue Aug 15 2006 Jakub Jelinek 2 4 90-21- if some test gets stuck, kill the tee process after make check finishe s - build with - mtune generic on i686 and x8664 Tue Aug 15 2006 Jakub Jelinek 2 4 90-20- PTHREADPRIOPROTECT support - fix errno if nice fails 201826 Thu Aug 10 2006 Jakub Jelinek 2 4 90-19- adaptive malloc brk mmap threshold - fix fchownat to use kernel syscall if available on many arches 201870 - only define ODIRECT with - DGNUSOURCE on ia64 to match all other arches 201748 Mon Aug 07 2006 Jakub Jelinek 2 4 90-18- NIS fixes - fix memusage and xtrace scripts 200736 - redirect sbin service sshd condrestart std to dev null when executed from glibcpostupgrade Wed Aug 02 2006 Jakub Jelinek 2 4 90-17- typo fix for the dladdr patch - build i 86 glibc with - mno-tls-direct-seg-refs 200469 Wed Aug 02 2006 Jakub Jelinek 2 4 90-16- fix dladdr on binaries libraries with only DTGNUHASH and no DTHASH 200635 - fix early timeout of initgroups data in nscd 173019 - add am pm display to esPE and esNI locales 167101 - fix nsscompat failures when nis nis unavailable 192072 Mon Jul 31 2006 Roland McGrath 2 4 90- 15- fix missing destructor calls in dlclose 197932 - enable transliteration support in all locales 196713 - disallow RTLDGLOBAL flag for dlmopen in secondary namespaces 197462 - PI mutex support Mon Jul 10 2006 Jakub Jelinek 2 4 90-13- DTGNUHASH support Fri Jun 30 2006 Jakub Jelinek 2 4 90-12- buildrequire gettext - enable fstatat64 newfstatat syscalls even on ppc s390 ia64 196494 - fix out of memory behavior in gettext 194321 - fix regex on multi-byte non-UTF-8 charsets 193873 - minor NIS fixes 190803 - don t use cancellable calls in posixspawn and only set id current thread if requested 193631 Wed May 31 2006 Jakub Jelinek 2 4 90-11- don t exit from nscd - i before the database is actually invalidated, add locking to prunecache 191464 - build glibc-devel i386 static libraries with - mno-tls-direct-seg-refs - DNOTLSDIRECTSEGREFS - RFC3542 support advanced API for IPv6 191001, BZ 2693 Wed May 24 2006 Jakub Jelinek 2 4 90-10- on i686 make glibc owner of lib i686 directory 192597 - search par ent NIS domains 190803 Sun May 21 2006 Jakub Jelinek 2 4 90-9- update from CVS - big NIS changes Fri May 19 2006 Jakub Jelinek 2 4 90-8- update from CVS - fix nsscompat when SETENTBATCHREAD TRUE is in etc default nss - fix RFC3484 precedence table for site-local and ULA addresses 188364 - fix a sunrpc memory leak Thu May 11 2006 Jakub Jelinek 2 4 90-7- update from CVS - fix tcgetattr 177965 - fix 191264 Fri May 05 2006 Jakub Jelinek 2 4 90-6- update from CVS - rebuilt using fixed rpm Fri May 05 2006 Jakub Jelinek 2 4 90-5- update from CVS - some NIS fixes - allow overriding rfc3484 address sorting tables for getaddrinfo through sample config file included in doc directory Mon May 01 2006 Jakub Jelinek 2 4 90-4- update from CVS - SETENTBATCHREAD etc default nss option for speeding up some usages of NIS 188246 - move debug state change notification 179208 - fix ldd script if one of the dynamic linkers is not installed 190259 Thu Apr 27 2006 Jakub Jelinek 2 4 90-3- update from CVS - fix a typo in 190085 - fix handling of SIGHUP in nscd when some caches are disabled 189978 - make nscd paranoia mode working with non-root server-user 189779 Wed Apr 26 2006 Jakub Jelinek 2 4 90-2- update from CVS - fix getaddrinfo 190002 - add auto-propagate options 177154 - fix nscd auditing 169148 Tue Apr 25 2006 Jakub Jelinek 2 4 90-1- update from CVS Mon Apr 24 2006 Jakub Jelinek 2 4-6- update from CVS - NIS fixes - don t segfault on too large argp key values 189545 - getaddrinfo fixes for RFC3484 188364 Tue Mar 28 2006 Jakub Jelinek 2 4-5- update from CVS - pshared robust mutex support - fix btowc and bwtoc in C 186410 - fix NIS 186592 - don t declare wcsto linternal for non-GCC or if not - O1 185667 - don t mention nscd failures on 2 0 kernels 185335 Tue Mar 07 2006 Roland McGrath 2 4-4- back up gdb conflicts to 2 4-3- really fix rintl on ppc64 Tue Mar 07 2006 Jakub Jelinek 2 4-2- accurate unwind info for lowlevellock h stubs on - fix ppc ppc64 ceill, floorl, rintl, roundl and truncl B Z 2423 Mon Mar 06 2006 Jakub Jelinek 2 4-1- update from CVS - glibc 2 4 release Mon Mar 06 2006 Jakub Jelinek 2 3 91-2- update from CVS - fix sYSMALLOc for MALLOCALIGNMENT 2 SIZESZ 183895 - revert ppc32 malloc alignment patch, it breaks mallocsetstate and needs some further thoughts and time 183894 - provide accurate unwind info for lowlevellock h stubs on x8664 Thu Mar 02 2006 Jakub Jelinek 2 3 91-1- update from CVS - fixes for various arches - ensure malloc returns pointers aligned to at least MIN 2 sizeof sizet , alignof long double only on ppc32 this has not been the case lately with addition of 128-bit long double, 182742 Wed Mar 01 2006 Jakub Jelinek 2 3 90-39- update from CVS Fri Feb 17 2006 Jakub Jelinek 2 3 90-38- update from CVS - robust mutexes rewrite Mon Feb 13 2006 Jakub Jelinek 2 3 90-37- update from CVS - at fixes - unshare syscall wrapper Sat Feb 04 2006 Jakub Jelinek 2 3 90-36- update from CVS - fix frequency setting for ITIMERPROF 179938, BZ 2268 - fix powerpc inline fegetround - fix nptldb 179946 Fri Feb 03 2006 Jakub Jelinek 2 3 90-35- update from CVS - handle futimesat fd, NULL, tvp as futimes fd, tvp - fix q cvt for - mlong-double-64 Thu Feb 02 2006 Jakub Jelinek 2 3 90-34- fix with C and - mlong-double-64 179742 - add nexttowardl redirect for - mlong-double-64 Thu Feb 02 2006 Jakub Jelinek 2 3 90-33- update from CVS - long double support fixes Wed Feb 01 2006 Jakub Jelinek 2 3 90-32- update from CVS - 128-bit long double fixes for ppc , s390 and sparc , alpha 128-bit long double support - add inotify syscall numbers to the override headers 179366 Mon Jan 30 2006 Jakub Jelinek 2 3 90-31- update from CVS - 128-bit long double on ppc, ppc64, s390, s390x and sparc - add some new syscall numbers to the override headers Mon Jan 09 2006 Jakub Jelinek 2 3 90-30- update from CVS. initializer fixes for - std c 9 on 32-bit arches - avoid writable 177121 Fri Jan 06 2006 Jakub Jelinek 2 3 90-29- update from CVS - make pthreadmutext an unnamed union again, as it a ffects libstdc ABI mangling Fri Jan 06 2006 Jakub Jelinek 2 3 90-28- update from CVS - make aiosuspend interruptible by signals 171968 Fri Jan 06 2006 Jakub Jelinek 2 3 90-27- only rely on dtype in 32-bit getdents on s390 for 2 6 11 Wed Jan 04 2006 Jakub Jelinek 2 3 90-26- update from CVS - for newly linked liolistio callers, send per request notifications 170116 - fixup nscd - S option removal changes 176860 - remove nonnull attribute from ctermid 176753 - fix PTHREAD INITIALIZER on 64-bit arches - SPARC NPTL support for pre-v9 CPUs - drop support for and 2 3 90-25- update from CVS - s390 and sparc pointer mangling fixes - install a sanitized LinuxThreads Mon Jan 02 2006 Jakub Jelinek 2 3 90-24- update from CVS - nscd audit changes 174422 - ppc vDSO support and ppc32 hp-timing Tue Dec 27 2005 Jakub Jelinek 2 3 90-23- update from CVS - robust mutexes - fix transliteration segfaults 176573, 176583 - ignore prelink temporaries in ldconfig 176570 Wed Dec 21 2005 Jakub Jelinek 2 3 90-22- updat e from CVS - minor fts fixes - revert broken Pragma workaround - fix ldconfig on bi-arch architectures 176316 Tue Dec 20 2005 Jakub Jelinek 2 3 90-21- update from CVS - fix pointer de mangling in gconvcache c Tue Dec 20 2005 Jakub Jelinek 2 3 90-20- update from CVS - time void 1 should segfault, not return - EFAULT 174856, BZ 1952 - fix errlist generation - update ulps for GCC 4 1 on IA-64 Mon Dec 19 2005 Jakub Jelinek 2 3 90-19- update from CVS - sysdeps generic reorg - setjmp longjmp jump pointer mangling - rebuilt with GCC 4 1-RH prerelease, worked around broken Pragma handling in it - remove glibc-profile subpackage - use non-PLT calls for malloc free realloc memalign invocations in mtrace and mcheck hooks 175261 - setjmp longjmp jump pointer mangling on ppc ia64 s390 Sat Nov 19 2005 Jakub Jelinek 2 3 90-18- update from CVS - change for broken apps that define const , handle non-GCC compilers - fix ppc strncmp BZ 1877, 173643, IT 83510 - provide shmattt typedef in ia64 2 3 90-17- update f rom CVS - fix in C - at fixes - epollwait is now a cancellation point Tue Nov 15 2005 Jakub Jelinek 2 3 90-16- update from CVS - make sure waitid syscall is used on ppc s390 Thu Oct 20 2005 Jakub Jelinek 2 3 90-15- update from CVS - be permissive in n check because of kernel bug 165351 171240 - don t misalign stack in pthreadonce on x8664 170786, IT 81521 - many locale fixes Mon Oct 10 2005 Jakub Jelinek 2 3 90-14- update from CVS - fix malloc bug after fork introduced in the last update - fix getent hosts IP for IPv4 IPs 169831 Mon Oct 03 2005 Jakub Jelinek 2 3 90-13- update from CVS - fix setuid etc hangs if some thread exits during the call 167766 - fix innetgr memory leak 169051 - support 2GB nscd log files 168851 - too many other changes to list here - include errno in nscd message if auditopen failed 169148 Mon Sep 12 2005 Jakub Jelinek 2 3 90-12- update from CVS - netgrp handling fixes 167728 - fix memory leak in setlocale BZ 1318 - fix hwcaps computation - several regex portabili ty improvements 167019 - hypotf fix - fix printf return code if underlying write fails BZ 1146 - PPC64 dl sym fixes for new ABI symbols - fix calloc with MALLOCPERTURB in environment on 64-bit architectures 166719 - source etc sysconfig nscd if it exists in etc rc d init d nscd 167083 - add triggerin for tzdata to glibc-common, so that tzdata updates update etc localtime and var spool postfix etc localtime if they exist 167787 Mon Aug 29 2005 Jakub Jelinek 2 3 90-11- FUTEXWAKEOP support to speed up pthreadcondsignal Wed Aug 24 2005 Jakub Jelinek 2 3 90-10- update from CVS - fix growing of nscd persistent database BZ 1204 - fix FORTIFYSOURCE mbstowcs and wcstombs if destination size is known at compile time, but length argument is not Mon Aug 22 2005 Jakub Jelinek 2 3 90-9- update from CVS - fix resolving over TCP 161181, 165802 - on ia64 don t abort on unhandled math function exception codes 165693 Mon Aug 08 2005 Jakub Jelinek 2 3 90-8- update from CVS - nscd persistent database verifi er 164001 - cleanup FORTIFYSOURCE bits h headers 165000 - handle EINTR in sigwait properly - make sure poor man s stack guard randomization keeps first byte 0 even on big-endian 32-bit arches - fix tst-stackguard1- obsolete linuxthreads-devel in glibc-devel Fri Jul 29 2005 Jakub Jelinek 2 3 90-7- update from CVS - do some poor man s stack guard randomization even without the costly --enable-stackguard-randomization - rebuilt with new GCC to make it use - msecure-plt on PPC32 Mon Jul 25 2005 Jakub Jelinek 2 3 90-6- update from CVS - fix execvp if PATH is not in environment and the call is going to fail BZ 1125 - another bits wchar2 h fix 163990 Fri Jul 22 2005 Jakub Jelinek 2 3 90-5- update from CVS - fix stubs h generation - don t use Gvalist in bits wchar2 h Fri Jul 22 2005 Jakub Jelinek 2 3 90-4- update from CVS - make sure bits wchar2 h header is installed - fix getgroupschk return type Thu Jul 21 2005 Jakub Jelinek 2 3 90-3- update from CVS - make sure nscd cmsg buffers aren t misaligned , handle EINTR from poll when contacting nscd more gracefully - remove malloc attribute from posixmemalign - correctly size nscd buffer for grpcache key 163538 - fix atan2f - fix error memory leaks - some more FORTIFYSOURCE protection Fri Jul 08 2005 Jakub Jelinek 2 3 90-2- update from CVS - ia64 stack protector support - handle DNS referral results as server errors 162625 - ctan fixes 160759 - pass argc, argv and envp also to executable s niarray functions BZ 974 - add ellipsis to clone prototype 161593 - fix glibc-profile 162601 - nsscompat fixes - use sysdeps generic version of in installed headers instead of NPTL version 162634 Mon Jun 27 2005 Jakub Jelinek 2 3 90-1- update from CVS - stack protector support - fix xdr on 64-bit arches 161583 - enable AATTGLIBC2 4 symbols - remove linuxthreads Mon Jun 20 2005 Jakub Jelinek 2 3 5-11- update from CVS - PPC32 - msecure-plt support - support classes keyword in 150350 - add RLIMITNICE and RLIMITRTPRIO to 157049 - decrease number of relocati ons in - use - laudit in nscd 159217 - handle big amounts of networking interfaces in getifaddrs ifnameindex 159399 - fix paIN locale s ampm 158715, BZ 622 - fix debugging of PIEs Mon May 30 2005 Jakub Jelinek 2 3 5-10- fix LDASSUMEKERNEL since 2 3 5-8 GLRO dlosversion has been always overwritten with the version of currently running kernel - remove linuxthreads man pages other than those covered in 3p section, as 3p man pages are far better quality and describe POSIX behaviour that NPTL implements 159084 Tue May 24 2005 Jakub Jelinek 2 3 5-9- update from CVS - increase bindresvport s LOWPORT to 512, apparently some broken daemons don t think 0 511 ports are reserved Mon May 23 2005 Jakub Jelinek 2 3 5-8- update from CVS - fix kernel version check in fix sendfile prototypes BZ 961 - try more ports in bindresvport if all 600 1023 are used, don t use priviledged ports when talking to portmap 141773 Fri May 20 2005 Jakub Jelinek 2 3 5-7- update from CVS - make regexec thread safe BZ 934 - fix statically linked programs on i 86, x8664, s390 and sparc 158027 - fix IBM939 iconv module BZ 955 Wed May 04 2005 Jakub Jelinek 2 3 5-6- update from CVS - fix cancellation on i 86 - add call frame information to i 86 assembly Tue May 03 2005 Jakub Jelinek 2 3 5-5- update from CVS - add some more UTF-8 locales 156115 - clean up lib64 tls instead of lib tls on x86-64, s390x and ppc64 in glibcpostupgrade 156656 - fix posixfallocate 156289 Thu Apr 28 2005 Jakub Jelinek 2 3 5-4- update from CVS - fix nscd cache pruning 150748 Wed Apr 27 2005 Jakub Jelinek 2 3 5-3- update from CVS - fix linuxthreads clocks - put xen libs into the glibc-2 i686 package instead of a separate one - fix symlink in linuxthreads-devel - do not include linuxthreads-devel on , just on the base architectures Wed Apr 27 2005 Jakub Jelinek 2 3 5-2- update from CVS - with MALLOCCHECK N N 0 153003 - fix recursive dlclose 154641 - handle z in strptime 154804 - automatically append obsolete linuxthreads to standard library search path if LDASSUMEKERNEL N N - glibc-xen subpackage for i686 Fri Apr 15 2005 Roland McGrath 2 3 5-1- update from CVS - fix execvp regression BZ 851 - ia64 libm updates - sparc updates - fix initstate strfry 154504 - grok PTNOTE in vDSO for kernel version and extra hwcap dirs, support hwcap keyword in files Mon Apr 04 2005 Jakub Jelinek 2 3 4-21- update from CVS - fix xdrrmtcallargs on 64-bit arches 151686 - fix. and with - std c89 - fexceptions 153774 Mon Apr 04 2005 Jakub Jelinek 2 3 4-20- move LinuxThreads libraries to obsolete linuxthreads and NPTL libraries to To run a program against LinuxThreads, LDLIBRARYPATH obsolete linuxthreads is now needed - bzip2 ChangeLog files instead of gzipping them Sat Apr 02 2005 Jakub Jelinek 2 3 4-19- update from CVS - fix nextafterl and several other libm routines on ia64 - fix initgroups BZ 661 - kill nptl-devel subpackage, add linuxthreads-devel, compile and link by default against NPTL and only with - I usr include linuxthreads - L usr linuxthr eads against LinuxThreads - package usr lib debug tls i 86 symlinks in i386 glibc-debuginfo - limit number of ChangeLog files in glibc-common doc to last 2 5 years of changes only to save space Fri Mar 25 2005 Jakub Jelinek 2 3 4-18- fix build on 64-bit arches with new GCC Thu Mar 24 2005 Jakub Jelinek 2 3 4-17- update from CVS - fix LDAUDIT in LinuxThreads - fix calloc with MPERTURB - fix error handling in pthreadcreate with PTHREADEXPLICITSCHED on ppc ia64 alpha mips BZ 801 - fix a typo in WINDOWS-31J charmap 151739 - fix NIS ypproterr 151469 Sun Mar 20 2005 Jakub Jelinek 2 3 4-16- fix pread with - DFILEOFFSETBITS 64 151573 Sat Mar 19 2005 Jakub Jelinek 2 3 4-15- update from CVS - better fix for the dlclose bug 145810, 150414 - fix regex crash on case insensitive search in zhCN locale 151215 - fix malloctrim BZ 779 - with - DFORTIFYSOURCE , avoid defining read and a bunch of others as function-like macros, there are too many broken programs out there - add dir gconv to glibc s file list 1 51372 Sun Mar 06 2005 Roland McGrath 2 3 4-14- fix bits socket2 h macro typos Sat Mar 05 2005 Jakub Jelinek 2 3 4-12- fix tst-chk - fix up ASNEEDED directive in usr BuildReq binutils 2 15 94 0 2-1 for ASNEEDED, in glibc-devel Conflict with binutils 2 3 4-11- update from CVS - fix execvp 149290 - fix dlclose 145810 - clear padding in 146614, BZ 776 - rebuilt with GCC4- changed GLIBCMINOR for now back to 3- back out the newly added GLIBC2 4 chk routines, instead do the checking in macros Sat Feb 12 2005 Jakub Jelinek 2 3 4-10- hopefully fix interaction with prelink 147655 Fri Feb 11 2005 Jakub Jelinek 2 3 4-9- update from CVS - bi-arch BZ 715 Fri Feb 11 2005 Jakub Jelinek 2 3 4-8- update from CVS - bi-arch BZ 632 - fix libdl on s390 and maybe other platforms - fix initstate BZ 710 - fix generation BZ 157 - define CMSPAR in bits termios h 147533 Tue Feb 08 2005 Jakub Jelinek 2 3 4-7- update from CVS - fix TLS handling in linuxthreads Tue Feb 08 2005 Jakub Jelinek 2 3 4-6- update from CVS - auditing - fix segfault if chrooted app attempts to dlopen a library and no standard library directory exists at all 147067, 144303 - fix initgroups when nscd is running, but has group caching disabled 146588 - fix pthreadkey in LinuxThreads when pthreadcreate has not been called yet 146710 - fix ppc64 swapcontext and setcontext 146736, BZ 700 - service nscd cosmetic fixes 146776 - fix IA-32 and x86-64 stack alignment in DSO constructors 145689 - fix zdump - v segfaults on x86-64 146210 - avoid calling sigaction SIGPIPE inside syslog 146021, IT 56686 - fix errno values for futimes BZ 633 - unconditionally include in malloc h BZ 650 - change regex B handling to match old GNU regex as well as perl grep s dfa from empty string inside of word to empty string not at a word boundary, BZ 693 - slightly optimize i686 TLS accesses, use direct TLS gs access in sem and allow building - mno-tls-direct-seg-refs glibc that is free of direct TLS gs access with negative offsets - fix addseverity - fix fmemopen - fix rewinddir - increase svc create listen backlog Thu Jan 06 2005 Jakub Jelinek 2 3 4-5- update from CVS - add some warnunusedresult marking - make ftruncate available even for just - DPOSIXCSOURCE 200112L BZ 640 Thu Jan 06 2005 Jakub Jelinek 2 3 4-4- update from CVS - fix IA-32 stack alignment for LinuxThreads thread functions and functions passed to clone 2 directly - fix ecvt on denormals 143279 - fix tlsgetaddr typo - fix rounding in IA-64 alarm 143710 - don t reinitialize environ in libcstartmain, so that effects of setenv putenv done in DSO initializers are preserved 144037, IT 57403 - fix fmemopen - fix vDSO lmapend and ltextend values - IA64 libm update 142494 - fix ppc rint ceil etc BZ 602 Tue Dec 21 2004 Jakub Jelinek 2 3 4-3- rebuilt Mon Dec 20 2004 Jakub Jelinek 2 3 4-2- work around rpm bug some more, this time by copying iconvconfig to iconvconfig Mon Dec 20 2004 Jakub Jelinek 2 3 4-1- update from CVS - glibc 2 3 4 release - add - o and --nostdlib options to ico nvconfig - if sbin ldconfig doesn t exist when running glibcpostupgrade , just don t attempt to run it This can happen during first install of bi-arch glibc and the other arch glibc s post wil run sbin ldconfig 143326 - use - o and --nostdlib options to create all needed files on bi-arch setups Sun Dec 19 2004 Jakub Jelinek 2 3 3-99- rebuilt Sat Dec 18 2004 Jakub Jelinek 2 3 3-98- add to glibcpostupgrade, only run telinit u if sbin init is the same ELF class and machine as glibcpostupgrade and similarly with condrestarting sshd 143046 Fri Dec 17 2004 Jakub Jelinek 2 3 3-97- update from CVS - fix ppc64 getcontext and swapcontext BZ 610 - sparc sparc64 fixes Wed Dec 15 2004 Jakub Jelinek 2 3 3-96- update from CVS - fix i686 USESTRINGINLINES strncat - make sure ppc ppc64 maintain correct stack alignment across clone Wed Dec 15 2004 Jakub Jelinek 2 3 3-95- export nisdomainofr from again which was unintentionally lost Wed Dec 15 2004 Jakub Jelinek 2 3 3-93- update from CVS - ppc ppc64 clone w ithout CLONETHREAD getpid adjustement - fix MALLOCCHECK for non-contiguous main arena BZ 457 - fix sysconf POSIXV6 for other ABI environments in bi-arch setups - s390 s390x clone without CLONETHREAD getpid adjustement Tue Dec 14 2004 Jakub Jelinek 2 3 3-92- update from CVS - fix libexec getconf filenames generation Tue Dec 14 2004 Jakub Jelinek 2 3 3-91- update from CVS - double buffer size in getXXbyYY or getXXent on ERANGE instead of adding BUFLEN 142617 - avoid busy loop in malloc if another thread is doing fork 142214 - some more realloc corruption checks - fix getconf POSIXV6WIDTHRESTRICTEDENVS output, tweak libexec getconf filenames Fri Dec 10 2004 Jakub Jelinek 2 3 3-90- update from CVS - regex speedups - use cat in ldd if running under bash3 to allow running it on binaries that are not through SELinux allowed to access console or tty - add NRwaitid defines for alpha and ia64 Wed Dec 08 2004 Jakub Jelinek 2 3 3-89- update from CVS - fix clone2 on ia64 - avoid tst-timer5 failing wit h linuxthreads implementation - if libcenablesecure, disallow mode normal - change ldd script to imply - r when - u is used, properly propagate return value and handle suid binaries Tue Dec 07 2004 Jakub Jelinek 2 3 3-88- update from CVS - disregard LDSHOWAUXV and LDDYNAMICWEAK if libcenablesecure - disregard LDDEBUG if libcenablesecure in normal mode if suid-debug doesn t exist - fix fseekpos after ungetc - avoid reading bytes before start of buffers in regex s checkdstlimitscalcpos1 142060 - make getpid working with clone clone2 without CLONETHREAD so far on i386 x8664 ia64 only - move libexec getconf to glibc from glibc-common - make libexec getconf directory owned by glibc package Fri Dec 03 2004 Jakub Jelinek 2 3 3-87- update from CVS - build libpthreadnonshared a objects with - fPIC on s390 s390x - fix mktime with 59 tmsec on entry - remove nonnull attribute for realpath - add make-target-directory for errlist-compat c rule hopefully fix 141404 - add testcase for ungetc bug - define POS IX CPUTIME to 0 on all Linux arches Tue Nov 30 2004 Jakub Jelinek 2 3 3-86- update from CVS - some posixopt h fixes - fix strtold use of unitialized memory 141000 - some more bugfixes for bugs detected by valgrind - rebuilt with GCC 3 4 3-5 to avoid packed stack layout on s390 139678 Fri Nov 26 2004 Jakub Jelinek 2 3 3-85- update from CVS - support - v specification in getconf - fix sysconf SCLFS64CFLAGS etc - avoid thread stack aliasing issues on EM64T 140803 - move include nptl headers from nptl-devel to glibc-headers, so that even NPTL specific programs can be built bi-arch without problems Wed Nov 24 2004 Jakub Jelinek 2 3 3-84- update from CVS - fix memory leak in getaddrinfo if using nscd 139559 - handle large lines in etc hosts and etc networks 140378 - add nonnull attributes to selected dirent h and dlfcn h functions Sun Nov 21 2004 Jakub Jelinek 2 3 3-83- update from CVS - add deprecated and or nonnull attribute to some signal h functions - speed up tzset by only using stat inste ad of open fstat when calling tzset for the second and following time if etc localtime has not changed - fix tgamma BZ 552 Sat Nov 20 2004 Jakub Jelinek 2 3 3-82- update from CVS - some malloc checking - libpthread a object dependency cleanups 115157 - fix for - std c89 - pedantic-errors 140132 Fri Nov 19 2004 Jakub Jelinek 2 3 3-81- don t use chunksize in 2 3 3-80- update from CVS - with - DFORTIFYSOURCE 2, prevent missing N formats - for - DFORTIFYSOURCE 2 and n in writable format string, issue special error message instead of using the buffer overflow detected one - speedup regex searching with REGNOSUB, add RENOSUB, speedup searching with nested subexps BZ 544 - block SIGCANCEL in NPTL timer helper thread - further free checking Tue Nov 16 2004 Jakub Jelinek 2 3 3-79- update from CVS - fix free checking - move etc default nss into glibc-common hopefully fix 132392 Mon Nov 15 2004 Jakub Jelinek 2 3 3-78- update from CVS - fix LDDEBUG statistics - issue error message before aborting in chkfa il - some more free checking Fri Nov 12 2004 Jakub Jelinek 2 3 3-77- update from CVS - speedup regex on palindromes BZ 429 - fix NPTL set ug id, so that even if making process less priviledged all threads change their credentials successfully Wed Nov 10 2004 Jakub Jelinek 2 3 3-76- update from CVS - fix regcomp crash 138439 - fix ftell 137885 - robustification of nscd to cope with corrupt databases 137140 - fix NPTL with pthreadexit immediately after pthreadcreate BZ 530 - some regex optimizations Tue Nov 02 2004 Jakub Jelinek 2 3 3-75- update from CVS - mktime cleanups BZ 487, 473 - unique comments in free 3 check error messages - adjust some x8664 headers for - m32 129712 - object size checking support even with GCC-3 4 2-RH 3 4 2-8 Wed Oct 27 2004 Jakub Jelinek 2 3 3-74- fix header - fix globfree 137176 - fix exiting if there are dlmopened libraries in namespaces other than main one not closed yet - export again resopcodes and p syms from that were lost in -69 Thu Oct 21 2004 Jakub Jeli nek 2 3 3-73- remove setaltroot and key also from Versions - back out syserrlist changes Thu Oct 21 2004 Jakub Jelinek 2 3 3-72- back out setaltroot and key addition - fix severe x86-64 symbol versioning regressions that breaks e g java binaries Wed Oct 20 2004 Jakub Jelinek 2 3 3-71- update from CVS - fix minor catchsegv temp file handling vulnerability CAN-2004-0968, 136319 - add 4 new errno codes - setaltroot, key syscalls on some arches - export dldebugstateAATTGLIBCPRIVATE from again for gdb purpose - use inetpton to decide what is address and what is hostname in getent 135422 - change dladdr dladdr1, so that dlisaddr is the same kind of value as dlsym dlvsym return makes difference on ia64 hppa only - fix catchsegv script so that it works with both 32-bit and 64-bit programs on multi-arch platforms Tue Oct 19 2004 Jakub Jelinek 2 3 3-70- update from CVS - require newer selinux-policy 135978 - add dir for var run nscd and var db nscd and ghost files in it - conflict with gcc4 4 0 0-0 6 and earlier needs builtinobjectsize Mon Oct 18 2004 Jakub Jelinek 2 3 3-69- update from CVS - object size checking support - DFORTIFYSOURCE Thu Oct 14 2004 Jakub Jelinek 2 3 3-68- update from CVS - support for namespaces in the dynamic linker - fix dlclose BZ 77 - uses now backtrace to work on IA-64, x86-64 and s390 130254 Tue Oct 12 2004 Jakub Jelinek 2 3 3-67- update from CVS - use non-blocking sockets in resolver 135234 - reset pd - res options on thread exit, so that threads reusing cached stacks get resolver state properly initialized BZ 434 Wed Oct 06 2004 Jakub Jelinek 2 3 3-66- update from CVS - avoid using perl in the spec file, buildrequire sed 3 95 127671 - export TIMEOUTFACTOR 16- fix JMPBUFCFAUNWINDSADJ on s390 Tue Oct 05 2004 Jakub Jelinek 2 3 3-65- update from CVS - define POSIXTHREADPROCESSSHARED and POSIXCLOCKSELECTION to -1 in LinuxThreads - define POSIXCPUTIME and POSIXTHREADCPUTIME to 0 on i 86 ia64 and make sure sysconf SC CPUTIME returns correct value - if POSIXCLOC KSELECTION -1 in nscd, still try sysconf SCCLOCKSELECTION and if it returns true, dlopen and dlsym pthreadcondattrsetclock - build nscd with - z relro and - z now Mon Oct 04 2004 Jakub Jelinek 2 3 3-64- update from CVS - stop using builtinexpect in assert and assertperror 127606 - try to avoid too much VA fragmentation with malloc on flexmap layout 118574 - nscd robustification - change valloc to use debugging hooks 134385 - make glibcpostupgrade more verbose on errors Fergal Daly, Fri Oct 01 2004 Jakub Jelinek 2 3 3-63- update from CVS - fix nscdgetgrouplist - fix a typo in x8664 pthreadmutextimedwait fix Fri Oct 01 2004 Jakub Jelinek 2 3 3-62- update from CVS - fix NPTL pthreadmutextimedwait on i386 x8664 BZ 417 Thu Sep 30 2004 Jakub Jelinek 2 3 3-61- update from CVS - some nscd fixes 134193 - cache initgroups in nscd 132850 - reread etc localtime in tzset even if just mtime changed 133481 - fix glob 126460 - another getmyaddress fix Wed Sep 29 2004 Jakub Jelinek 2 3 3-60- update from C VS - fix getmyaddress 133982 - remove nonnull attribute from second utime argument 133866 - handle SIGSETXID the same way as SIGCANCEL in sigaction pthreadkill sigwait sigwaitinfo etc - add extension to long long types in NPTL Mon Sep 27 2004 Jakub Jelinek 2 3 3-59- update from CVS - fix BZ 151, 362, 381, 407 - fdim fix for inf inf BZ 376 Sun Sep 26 2004 Jakub Jelinek 2 3 3-58- update from CVS - vasprintf fix BZ 346 - gettext locking BZ 322 - change linuxthreads useldt h inclusion login again, the last one failed all linuxthreads FLOATINGSTACKS tests Sat Sep 25 2004 Jakub Jelinek 2 3 3-57- update from CVS - fix setuid in LDASSUMEKERNEL 2 2 5 libc 133558 - fix nis locking 132204 - RTLDDEEPBIND support - fix pthreadcreate bugs BZ 401, 405 Wed Sep 22 2004 Roland McGrath 2 3 3-56- migrated CVS to fedora-branch in glibc repository - source tarballs renamed - redhat moved to fedora , some old cruft removed - update from trunk - some nonnull annotations Wed Sep 22 2004 Jakub Jelinek 2 3 3-55- update from CVS - set ug id now affect the whole process in NPTL - return EAGAIN instead of ENOMEM when not enough memory in pthreadcreate Fri Sep 17 2004 Jakub Jelinek 2 3 3-54- update from CVS - nscd getaddrinfo caching Tue Sep 14 2004 Jakub Jelinek 2 3 3-53- restore temporarily old definition of P PMT for third party apps Tue Sep 14 2004 Jakub Jelinek 2 3 3-52- update from CVS - nscd bi-arch fix - remove all uses of P PMT from glibc headers - update and reenable nscd SELinux patch - remove libnss1 and libnss compatibility NSS modules on IA-32, SPARC and Alpha Fri Sep 10 2004 Jakub Jelinek 2 3 3-51- update from CVS - disable one of the malloc double free checks for non-contiguous arenas where it doesn t have to be true even for non-broken apps Thu Sep 09 2004 Jakub Jelinek 2 3 3-50- update from CVS - pwd grp host loops with nscd speed up by sharing the nscd cache r o with applications - inexpensive double free check in free 3 - make NPTL pthread h initializers usable even from C BZ 375 - use atomic instructions even in i386 nscd on i486 CPUs conditionally Fri Sep 03 2004 Jakub Jelinek 2 3 3-49- update from CVS - fix linuxthreads tst-cancel Fri Sep 03 2004 Jakub Jelinek 2 3 3-48- update from CVS - fix pthreadconddestroy BZ 342 - fix fnmatch without FNMNOESCAPE BZ 361 - fix ppc32 setcontext BZ 357 - add NPTL support for i386 glibc only if run on i486 or higher CPU - add NRwaitid defines for i386, x8664 and sparc Tue Aug 31 2004 Jakub Jelinek 2 3 3-47- update from CVS - persistent nscd caching - ppc64 32-bit atomicity fix - fix x86-64 nptl-devel headers for - m32 compilation - ghost 130597 - edit in glibcpostupgrade if include line is missing 120588 - ugly hacks for the IA-64 emul braindamage 124996, 128267 Sat Aug 21 2004 Jakub Jelinek 2 3 3-46- update from CVS Thu Aug 19 2004 Jakub Jelinek 2 3 3-45- update from CVS - fix nsscompat s initgroups handling 130363 - fix getaddrinfo aicanonname setting Thu Aug 19 2004 Jakub Jelinek 2 3 3-44- update from CVS - add ip6-dotint o ption, make no-ip6-dotint the default - BuildPrereq libselinux-devel 129946 - on ppc64, build without dot symbols Thu Aug 12 2004 Jakub Jelinek 2 3 3-43- update from CVS - remove debugging printout 129747 - make usable in C IT 45148 - update RLIMIT constants in make POSIX compliant 129740 Wed Aug 11 2004 Jakub Jelinek 2 3 3-42- fix last tzset fixes, disable rereading of etc localtime every time for now - really enable SELinux support for NSCD Wed Aug 11 2004 Jakub Jelinek 2 3 3-41- update from CVS - freadunlocked fwriteunlocked macro fixes BZ 309, 316 - tzset fixes BZ 154 - speed up pthreadrwlockunlock on arches other than i386 and x8664 129455 - fix compilation with - ansi resp - std c89 or - std c99 and - DXOPENSOURCE 56 00 but no - DPOSIXSOURCE or - DPOSIXCSOURCE BZ 284 - add SELinux support for NSCD Fri Aug 06 2004 Jakub Jelinek 2 3 3-40- update from CVS - change resinit to force all threads to re-initialize resolver before they use it next time 125712 - various getaddrinfo and related fix es BZ 295, 296 - fix IBM iconv modules 128674 - some nscd fixes e g BZ 292 - RFC 3678 support Multicast Source Filters - handle lib i686 librtkaio - in i386 glibcpostupgrade the same as lib i686 librt - Fri Jul 23 2004 Jakub Jelinek 2 3 3-39- update from CVS - conformance related changes in headers - remove - finline-limit 2000 for GCC 3 4 x Thu Jul 22 2004 Jakub Jelinek 2 3 3-38- update from CVS - fix resinit leaks - fix newlocale races - fix ppc64 setjmp - fix strtold BZ 274 Fri Jul 16 2004 Jakub Jelinek 2 3 3-37- update from CVS - allow pthreadcancel in DSO destructors run at exit time - fix pow on IA-32 and powl on x86-64- allow PIEs on IA-32 to have main in a shared library they depend on Mon Jul 05 2004 Jakub Jelinek 2 3 3-36- s390 slot reduction - fix pthreadrwlocktimedrdlock on x8664 Wed Jun 30 2004 Jakub Jelinek 2 3 3-35- tweak spec file for the - NPTL changes Wed Jun 30 2004 Jakub Jelinek 2 3 3-34- update from CVS - ifnameindex using preferably netlink - printfparsemb initialization fix - NPTL version is now the same as glibc version Mon Jun 28 2004 Jakub Jelinek 2 3 3-33- update from CVS - reread for nscd --invalidate hosts - fix FGETLK FSETLK FSETLKW constants on x8664 for - m32 - DFILEOFFSETBITS 64 compilations - avoid calling non-existing fcntl64 syscall on ppc64 Mon Jun 14 2004 Jakub Jelinek 2 3 3-32- update from CVS - FUTEXCMPREQUEUE support fix pthreadcond deadlocks - fix backtrace in statically linked programs - rebuilt with GCC 3 4, adjusted ulps and i386 Fri May 28 2004 Jakub Jelinek 2 3 3-31- update from CVS - and changes for GCC 3 - make cstubs buildable even with GCC 3 2 x 123042 Fri May 21 2004 Jakub Jelinek 2 3 3-30- fix pthreadcondwait on architectures other than IA-32 and x8664 Thu May 20 2004 Jakub Jelinek 2 3 3-29- use lib64 instead of lib on ia64 if is defined to lib64 Wed May 19 2004 Jakub Jelinek 2 3 3-28- update from CVS - FUTEXREQUEUE fixes 115349 - SPARC GCC 3 4 build fix - fix handling of undefined TLS symbols on IA32 RELA only , SPARC and S H - regex translate fix - speed up sprintf - x8664 makecontext alignment fix - make POSIX sigpause the default sigpause, unless BSD sigpause requested Tue May 11 2004 Jakub Jelinek 2 3 3-27- remove in glibcpostupgrade on x86-64, s390x and ppc64 instead of build mq with - fexceptions Fri May 07 2004 Jakub Jelinek 2 3 3-26- update from CVS - fix - fix memory leaks in nis, getifaddrs, etc caused by incorrect use of realloc - remove lib in glibcpostupgrade and rerun ldconfig if needed, otherwise after glibc upgrade might be a stale symlink Wed May 05 2004 Jakub Jelinek 2 3 3-25- update from CVS - disable FUTEXREQUEUE work around 115349 - mq for sparc sparc64 ia64 Tue May 04 2004 Jakub Jelinek 2 3 3-24- update from CVS - define SISSOCK in - DXOPENSOURCE 600 and SI FS SOCK plus F SG ETOWN also in - DXOPENSOURCE 500 both included already in XNS5 - reorder dlopen checks, so that dlopening ETREL objects complains about ETDYN ETEXEC, not about phentsize 121606 - fix strpbrk macro for GCC 3 4 BZ 130 - fix BZ 140 - sched gs etaffinity documentation fix BZ 131 - fix sparc64 build BZ 139 - change linuxthreads back to use non-cancellable writes to manager pipes etc - fix semtimedwait return value in linuxthreads BZ 133 - ia64 unnecessary PLT relocs removal Thu Apr 22 2004 Jakub Jelinek 2 3 3-23- update from CVS - fix scanf - fix shmunlink, semunlink and mqunlink errno values - avoid memory leaks in error - execstack fixes on s390 Mon Apr 19 2004 Jakub Jelinek 2 3 3-22- update from CVS - mq and timer fixes - rebuilt with binutils 2 15 90 0 3-2 to fix IA-64 statically linked binaries - fix linuxthreads on s390 , so it is no longer DTTEXTREL Sat Apr 17 2004 Jakub Jelinek 2 3 3-21- disable rtkaio - update from CVS - POSIX message passing support - fixed SIGEVTHREAD support for POSIX timers - fix free on non-malloced memory in syslog - fix ffsl on some 64-bit arches - fix schedsetaffinity on x86-64, ia64 - fix ppc64 umount - NETIDAUTHORITATIVE, SERVICESAUTHORITATIVE support - various NIS speed ups - fix fwrite with 2GB sizes on 64-bit arches - fix pthreadgetattrnp guardsize reporting in NPTL - report PLT relocations in and during the build Thu Mar 25 2004 Jakub Jelinek 2 3 3-20- update from CVS - change NPTL PTHREADMUTEXADAPTIVENP mutexes to spin on SMP - strtol speed optimization - don t try to use certainly unimplemented syscalls on ppc64- kill - debug subpackage, move the libs to glibc-debuginfo into usr lib debug usr directory - fix cstubs with gcc 3 4- move all the up to 3 builds into build scriptlet and leave only installation in the install scriptlet Mon Mar 22 2004 Jakub Jelinek 2 3 3-19- update from CVS - affinity API changes Thu Mar 18 2004 Jakub Jelinek 2 3 3-18- update from CVS - fix ia64 iopl 118591 - add support for - fix x86-64 LDDEBUG statistics - fix hwcap handling when using 118518 Mon Mar 15 2004 Jakub Jelinek 2 3 3-17- update from CVS - implement non-l function on top of l functions Thu Mar 11 2004 Jakub Jelinek 2 3 3-16- update from CVS - fix s390 TLS handlin g Wed Mar 10 2004 Jakub Jelinek 2 3 3-15- update from CVS - special section for compatibility code - make getpid work even in vfork child - configure with --enable-bind-now to avoid lazy binding in and Fri Mar 05 2004 Jakub Jelinek 2 3 3-14- update from CVS - fix iconv - c 117021 - fix PIEs on sparc sparc64 - fix posixfadvise on 64-bit architectures - add locale-archive as ghost file 117014 Mon Mar 01 2004 Jakub Jelinek 2 3 3-13- update from CVS Fri Feb 27 2004 Jakub Jelinek 2 3 3-12- update from CVS Fri Feb 27 2004 Jakub Jelinek 2 3 3-11- update from CVS - fix when vDSO is randomized Fri Feb 20 2004 Jakub Jelinek 2 3 3-10- update from CVS Fri Feb 20 2004 Jakub Jelinek 2 3 3-9- update from CVS Tue Feb 10 2004 Jakub Jelinek 2 3 3-8- update from CVS Tue Jan 27 2004 Jakub Jelinek 2 3 3-7- update from CVS - dliteratephdr extension to signal number of added removed libraries - fix PTGNURELRO support on ppc with prelinking Fri Jan 23 2004 Jakub Jelinek 2 3 3-6- rebuilt with fixed GCC on IA-64 Th u Jan 22 2004 Jakub Jelinek 2 3 3-5- fix PTGNURELRO support Wed Jan 21 2004 Jakub Jelinek 2 3 3-4- update from CVS - some further regex speedups - fix handling in regex 112869 - change regfree to match old regex behaviour what is freed and clearing of freed pointers - fix accesses to unitialized memory in regex 113507, 113425, - PTGNURELRO support Tue Dec 30 2003 Jakub Jelinek 2 3 3-3- update from CVS - fix pmapset fd and memory leak 112726 - fix backreference handling in regex - rebuilt under glibc without the above bug to fix linker script 112738 Mon Dec 29 2003 Jakub Jelinek 2 3 3-2- update from CVS - faster getpid in NPTL builds - fix to make pthreadsetcancelstate PTHREADCANCELDISABLE, really disable cancellation 112512 - more regex fixes and speedups - fix nextafter nexttoward - handle 6th syscall 3 argument on AMD64 - handle memalign posixmemalign in mtrace - fix linuxthreads memory leak 112208 - remove throw from cancellation points in linuxthreads 112602 - fix NPTL unregisteratf ork - fix unwinding through alternate signal stacks Mon Dec 01 2003 Jakub Jelinek 2 3 3-1- update from CVS - 2 3 3 release - lots of regex fixes and speedups 110401 - fix atan2 - fix pshared condvars in NPTL - fix pthreadattrdestroy for attributes created with pthreadattrinitAATTGLIBC2 0- for the time being, include both nbNO and noNO as locales so that the distribution can catch up with the noNO - nbNO transition - add BuildPrereq texinfo 110252 Tue Nov 18 2003 Jakub Jelinek 2 3 2-102- update from CVS - fix getifaddrs CAN-2003-0859 - fix ftw fd leak - fix linuxthreads sigaction 108634 - fix glibc 2 0 stdio compatibility - fix uselocale LCGLOBALLOCALE - speed up stdio locking in non-threaded programs on IA-32 - try to maintain correct order of cleanups between those registered with attribute cleanup and with LinuxThreads style pthreadcleanuppush pop 108631 - fix segfault in regex 109606 - fix REICASE multi-byte handling in regex - fix pthreadexit in libpthread a 109790 - FTWACTIONRETVAL support - lots of regex fixes and speedups - fix ceill floorl on AMD64 Mon Oct 27 2003 Jakub Jelinek 2 3 2-101- update from CVS - fix --verify and ldd Mon Oct 27 2003 Jakub Jelinek 2 3 2-100- update from CVS - fix sprof 103727 - avoid infinite loops in statvfs with hosed mounts file - prevent dlopening of executables - fix glob with GLOBBRACE and without GLOBNOESCAPE - fix locale printing of word values on 64-bit big-endian arches 107846 - fix getnameinfo and getaddrinfo with reverse IPv6 lookups 101261 Wed Oct 22 2003 Jakub Jelinek 2 3 2-99- update from CVS - dliteratephdr in libc a on arches other than IA-64 - LDDEBUG statistics prints number of relative relocations - fix hwcap computation - NPTL is now part of upstream glibc CVS - include ZA locales Sat Oct 04 2003 Jakub Jelinek 2 3 2-98- update from CVS - fix close, pause and fsync 105348 - fix pthreadonce on IA-32- implement backtrace on IA-64, handle - fomit-frame-pointer in AMD64 backtrace 90402 Tue Sep 30 2003 Jakub Jelinek 2 3 2- 97- update from CVS - fix with C or - ansi or - pedantic C - fix mknod ustat return value when given bogus device number 105768 Fri Sep 26 2003 Jakub Jelinek 2 3 2-96- rebuilt Fri Sep 26 2003 Jakub Jelinek 2 3 2-95- fix IA-64 getcontext Thu Sep 25 2003 Jakub Jelinek 2 3 2-94- update from CVS - fix syslog with non-C non-en locales 61296, 104979 - filter GLIBCPRIVATE symbols from glibc provides - fix NIS Thu Sep 25 2003 Jakub Jelinek 2 3 2-93- update from CVS - assume 2 4 21 kernel features on RHEL ppc , so that context works - backout execstack support for RHEL - build rtkaio on amd64 too Wed Sep 24 2003 Jakub Jelinek 2 3 2-92- update from CVS - execstack noexecstack support - build nscd as PIE - move libcstackend back to AATTGLIBC2 1- build against elfutils 0 86 to fix stripping on s390x Mon Sep 22 2003 Jakub Jelinek 2 3 2-91- rebuilt Mon Sep 22 2003 Jakub Jelinek 2 3 2-90- update from CVS - NPTL locking change 102682 - don t jump around lock on amd64 Thu Sep 18 2003 Jakub Jelinek 2 3 2-89- fi x openmemstream syslog 104661 Thu Sep 18 2003 Jakub Jelinek 2 3 2-88- update from CVS - retrieve affinity in pthreadgetattrnp - fix pthreadattr gs etaffinitynp - handle hex and octal in wordexp Wed Sep 17 2003 Jakub Jelinek 2 3 2-87- update from CVS - truncate instead of round in utimes when utimes syscall is not available - don t align stack in every glibc function unnecessarily on IA-32 - make sure threads have their stack 16 byte aligned on IA-32 - move sched sg etaffinity to GLIBC2 3 3 symbol version 103231 - fix pthreadgetattrnp for the initial thread 102683 - avoid linuxthreads signal race 104368 - ensure all gzip invocations are done with - n option Fri Sep 12 2003 Jakub Jelinek 2 3 2-86- update from CVS - avoid linking in libgcceh a unnecessarily - change ssizet back to long int on s390 - m31, unless gcc 2 95 x is used Wed Sep 10 2003 Jakub Jelinek 2 3 2-85- update from CVS - fix IA-64 memccpy 104114 Tue Sep 09 2003 Jakub Jelinek 2 3 2-84- update from CVS - undo broken amd64 signal context changes Tue Sep 09 2003 Jakub Jelinek 2 3 2-83- update from CVS - change nlinkt, ssizet and intptrt types on s390 - m31 to int - change uquadt, quadt, qaddrt, devt, ino64t, lofft, off64t, rlim64t, blkcnt64t, fsblkcnt64t, fsfilcnt64t on 64-bit arches from long long int to long int to restore binary compatibility for C functions using these types as arguments Sun Sep 07 2003 Jakub Jelinek 2 3 2-82- rebuilt Sat Sep 06 2003 Jakub Jelinek 2 3 2-81- update from CVS - fix tc gs etattr cf gs et io speed on ppc 102732 - libio fixes Thu Sep 04 2003 Jakub Jelinek 2 3 2-80- update from CVS - fix IA-64 cancellation when mixing attribute cleanup and old-style pthreadcleanuppush cleanups Tue Sep 02 2003 Jakub Jelinek 2 3 2-79- updated from CVS - lots of cancellation fixes - fix posixfadvise on ppc32 - TLS layout fix - optimize stdio cleanups 103354 - sparcv9 NPTL - include sigset, sighold, sigrelse, sigpause and sigignore prototypes in signal h even if - DXOPENSOURCEEXTENDED 103269 - fix svcgetr eqset on 64-bit big-endian arches - return ENOSYS in linuxthreads pthreadbarrierattrsetpshared for PTHREADPROCESSSHARED - add pthreadcondtimedwait stubs to 102709 - split glibc-devel into glibc-devel and glibc-headers to ensure amd64 usr include always wins on amd64 i386 bi-arch installs - increase PTHREADSTACKMIN on alpha, ia64 and sparc - get rid of syscall prototypes and stubs in sysdeps unix sysv linux - run make check also with linuxthreads on IA-32 non-FLOATINGSTACKS and NPTL on IA-32 also FLOATINGSTACKS linuxthreads libraries and tests Mon Aug 25 2003 Jakub Jelinek 2 3 2-78- include dl-osinfo h only in glibc-debuginfo-2 not in glibc-debuginfo-common Mon Aug 25 2003 Jakub Jelinek 2 3 2-77- update from CVS - fix glibc 2 0 libio compatibility 101385 - fix ldconfig with usr lib lib symlinks 102853 - fix assert h 102916, 103017 - make identical between IA-32 and AMD64 102887 - fix static linking of large IA-64 binaries 102586 - avoid using floating point regs in lazy binding code on pp c64 102763 Fri Aug 22 2003 Roland McGrath 2 3 2-76- add tdthrtlsgetaddr changes missed in initial nptldb rewrite Sun Aug 17 2003 Roland McGrath 2 3 2-74- nptldb rewrite not yet in CVS Thu Aug 14 2003 Jakub Jelinek 2 3 2-72- update from CVS - fix rtkaio aiofsync - update rtkaio for BROKENTHREADSIGNALS - fix assert macro when used on pointers Wed Aug 13 2003 Jakub Jelinek 2 3 2-71- update from CVS Tue Aug 12 2003 Jakub Jelinek 2 3 2-70- update from CVS - disable CLONESTOPPED for now until it is resolved - strip crt files - fix libio on arches with no 2 3 2-69- update from CVS - fix pthreadcreateAATTGLIBC2 0 101767 - ASSUMECLONESTOPPED on all arches but s390 in RHEL Sun Aug 03 2003 Jakub Jelinek 2 3 2-68- update from CVS - only use CLONESTOPPED if kernel supports it, fix setting of thread explicit scheduling 101457 Fri Aug 01 2003 Jakub Jelinek 2 3 2-67- update from CVS - fix utimes and futimes if kernel doesn t support utimes syscall - fix s390 ssizet type - fix dlerror when called before a ny dlopen dlsym - update IA-64 bits sigcontext h 101344 - various warning fixes - fix pthread h comment typos 101363 Wed Jul 30 2003 Jakub Jelinek 2 3 2-66- update from CVS - fix dlopen of libraries using TLS IE LE models Tue Jul 29 2003 Jakub Jelinek 2 3 2-65- update from CVS - fix timercreate - use extension before long long typedefs in 100718 Mon Jul 28 2003 Jakub Jelinek 2 3 2-64- update from CVS - fix wcpncpy 99462 - export resAATTGLIBC2 0 even from NPTL resstate unlike errnolocation or herrnolocation was introduced in glibc 2 2 - fix zic bug on 64-bit platforms - some TLS handling fixes - make ldconfig look into alternate ABI dirs by default 99402 - move zoneinfo to tzdata package, so that it can be errataed separately from glibc - new add-on - rtkaio - prereq libgcc, as glibc now relies on for pthreadcancel Tue Jul 15 2003 Jakub Jelinek 2 3 2-63- fix thread cancellation on ppc64 Sat Jul 12 2003 Jakub Jelinek 2 3 2-62- update from CVS - fix thread cancellation on ppc32, s390 and s39 0x Thu Jul 10 2003 Jakub Jelinek 2 3 2-61- update from CVS - build libcnonshared a with - fPIC instead of - fpic - fix ppc64 PIE support - add cfi directives to NPTL sysdep-cancel h on ppc ppc64 s390 s390x Tue Jul 08 2003 Jakub Jelinek 2 3 2-60- update from CVS Thu Jul 03 2003 Jakub Jelinek 2 3 2-59- update from CVS - on IA-64 use different symbols for cancellation portion of syscall handlers to make gdb happier Thu Jun 26 2003 Jakub Jelinek 2 3 2-58- update from CVS - nsscompat supporting LDAP etc Tue Jun 24 2003 Jakub Jelinek 2 3 2-57- update from CVS Thu Jun 19 2003 Jakub Jelinek 2 3 2-56- fix condvars and semaphores in ppc NPTL - fix test-skeleton c reporting of timed-out tests 91269 - increase timeouts for tests during make check Wed Jun 18 2003 Jakub Jelinek 2 3 2-55- make ldconfig default to both lib usr lib and lib64 usr lib64 on bi-ABI architectures 97557 - disable FUTEXREQUEUE on ppc temporarily Wed Jun 18 2003 Jakub Jelinek 2 3 2-54- update from CVS - fix glibcpostupgrade on ppc Tu e Jun 17 2003 Jakub Jelinek 2 3 2-53- update from CVS - fix localedef 90659 - tweak linuxthreads for librt cancellation Mon Jun 16 2003 Jakub Jelinek 2 3 2-52- update from CVS Thu Jun 12 2003 Jakub Jelinek 2 3 2-51- update from CVS - fix 97169 Wed Jun 11 2003 Jakub Jelinek 2 3 2-50- update from CVS Tue Jun 10 2003 Jakub Jelinek 2 3 2-49- update from CVS - fix pthreadcondsignal on IA-32 92080, 92253 - fix setegid 91567 - don t prelink - R on any architecture, it prohibits address randomization Thu Jun 05 2003 Jakub Jelinek 2 3 2-48- update from CVS - fix IA-64 NPTL build Thu Jun 05 2003 Jakub Jelinek 2 3 2-47- update from CVS - PTGNUSTACK segment in binaries executables and section in oa Sun Jun 01 2003 Jakub Jelinek 2 3 2-46- update from CVS - enable NPTL on AMD64- avoid using trampolines in localedef Thu May 29 2003 Jakub Jelinek 2 3 2-45- enable NPTL on IA-64 Thu May 29 2003 Jakub Jelinek 2 3 2-44- update from CVS - enable NPTL on s390 and s390x - make initarraystart etc symbols in hidden u ndefined Thu May 29 2003 Jakub Jelinek 2 3 2-43- update from CVS Fri May 23 2003 Jakub Jelinek 2 3 2-42- update from CVS Tue May 20 2003 Jakub Jelinek 2 3 2-41- update from CVS - use NPTL libs if uname - r contains nptl substring or is 2 5 69 or settidaddress syscall is available instead of checking ATSYSINFO dynamic tag Thu May 15 2003 Jakub Jelinek 2 3 2-40- update from CVS Wed May 14 2003 Jakub Jelinek 2 3 2-39- update from CVS - fix for prelinking of libraries with no dependencies Tue May 13 2003 Jakub Jelinek 2 3 2-38- update from CVS - enable NPTL on ppc and ppc64 Tue May 06 2003 Matt Wilson 2 3 2-37- rebuild Sun May 04 2003 Jakub Jelinek 2 3 2-36- update from CVS Sat May 03 2003 Jakub Jelinek 2 3 2-35- update from CVS - make - jN build fixes Fri May 02 2003 Jakub Jelinek 2 3 2-34- update from CVS - avoid using trampolines in iconvconfig for now Sat Apr 26 2003 Jakub Jelinek 2 3 2-33- update from CVS Fri Apr 25 2003 Jakub Jelinek 2 3 2-32- update from CVS - more ppc TLS fixes Wed Apr 2 3 2003 Jakub Jelinek 2 3 2-31- update from CVS - nscd fixes - fix Bahrain spelling 56298 - fix Ukrainian collation 83973 - accept trailing spaces in 86032 - perror fix 85994 - fix localedef 88978 - fix getifaddrs 89026 - fix strxfrm 88409 - fix ppc TLS - fix getaddrinfo 89448 - don t print warning about errno, herrno or res if LDASSUMEKERNEL 2 4 1 or earlier Tue Apr 15 2003 Jakub Jelinek 2 3 2-30- update from CVS - fix prelink on ppc32- add TLS support on ppc32 and ppc64- make sure on - m64 arches all helper binaries are built with this option Mon Apr 14 2003 Jakub Jelinek 2 3 2-29- update from CVS - fix strxfrm 88409 - use - m64 - mno-minimal-toc on ppc64- conflict with kernels 2 3 2-28- update from CVS - fix NPTL pthreaddetach and already terminated, but not yet joined thread 88219 - fix bug-regex4 testcase 88118 - reenable prelink support broken in 2 3 2-13 - fix registerprintffunction 88052 - fix double free with fopen using ccs 88056 - fix potential access below esp in context 88093 - fix buffer underrun in gencat - H 88099 - avoid using unitialized variable in tst-tgmath 88101 - fix gammal 88104 - fix iconv - c - fix xdrstring PR libc 4999 - fix symlink - avoid running NPTL cleanups twice in some cases - unblock pthreadsignalcancel in linuxthreads, so that linuxthreads threaded programs work correctly if spawned from NPTL threaded programs - fix sysconf SC , PHYSPAGES - remove lib i686 directory before running ldconfig in glibc post during i686 - i386 glibc upgrades 88456 Wed Apr 02 2003 Jakub Jelinek 2 3 2-22- update from CVS - add pthreadatfork to libpthread a Tue Apr 01 2003 Jakub Jelinek 2 3 2-21- update from CVS - make sure linuxthreads pthreadmutexlock etc is not a cancellation point Sat Mar 29 2003 Jakub Jelinek 2 3 2-20- update from CVS - if kernel 2 4 1 doesn t support NPTL, fall back to lib i686 libs on i686, not stright to lib Fri Mar 28 2003 Jakub Jelinek 2 3 2-19- update from CVS - timers fixes Thu Mar 27 2003 Jakub Jelinek 2 3 2-18- update from CVS - fix N PTL pthreadcondtimedwait - fix sysconf SCMONOTONICCLOCK - use tls instead of lib tls on x86-64- add tls librt so and i686 librt so - display content of files for all make check failures Wed Mar 26 2003 Jakub Jelinek 2 3 2-17- update from CVS - kernel POSIX timers support Sat Mar 22 2003 Jakub Jelinek 2 3 2-16- update from CVS - export fork from glibc again - fix glibc-compat build in NPTL - fix cstubs - fix some more atomic h problems - don t check abi in glibc-compat libs Fri Mar 21 2003 Jakub Jelinek 2 3 2-15- update from CVS - build glibc-compat for glibc 2 0 compatibility and cstubs add-ons - condrestart sshd in glibcpostupgrade so that the user can log in remotely and handle the rest 86339 - fix a typo in glibcpostupgrade on sparc Tue Mar 18 2003 Jakub Jelinek 2 3 2-14- update from CVS - change i686 athlon base to 0x00e80000 Mon Mar 17 2003 Jakub Jelinek 2 3 2-13- update from CVS - hopefully last fix for condvar problems Fri Mar 14 2003 Jakub Jelinek 2 3 2-12- fix bits syscall h creation on x86-64 Thu Mar 13 2003 Jakub Jelinek 2 3 2-11- update from CVS Wed Mar 12 2003 Jakub Jelinek 2 3 2-10- update from CVS Tue Mar 11 2003 Jakub Jelinek 2 3 2-9- update from CVS - fix glibc-debug description 85111 - make a symlink again, not linker script Tue Mar 04 2003 Jakub Jelinek 2 3 2-8- update from CVS - remove the workarounds for broken software accessing GLIBCPRIVATE symbols Mon Mar 03 2003 Jakub Jelinek 2 3 2-7- update from CVS Sun Mar 02 2003 Jakub Jelinek 2 3 2-6- fix TLS IE LE model handling in dlopened libraries on TCBATTP arches Tue Feb 25 2003 Jakub Jelinek 2 3 2-5- update from CVS Tue Feb 25 2003 Jakub Jelinek 2 3 2-4- update from CVS Mon Feb 24 2003 Jakub Jelinek 2 3 2-3- update from CVS - only warn about errno, herrno or res for binaries, never libraries - rebuilt with gcc-3 2 2-4 to use direct gs TLS access insn sequences Sun Feb 23 2003 Jakub Jelinek 2 3 2-2- update from CVS Sat Feb 22 2003 Jakub Jelinek 2 3 2-1- update from CVS Thu Feb 20 2003 Jakub Jelinek 2 3 1-51- u pdate from CVS Wed Feb 19 2003 Jakub Jelinek 2 3 1-50- update from CVS Wed Feb 19 2003 Jakub Jelinek 2 3 1-49- update from CVS - remove nisplus and nis from the default 67401, 9952 Tue Feb 18 2003 Jakub Jelinek 2 3 1-48- update from CVS Sat Feb 15 2003 Jakub Jelinek 2 3 1-47- update from CVS Fri Feb 14 2003 Jakub Jelinek 2 3 1-46- update from CVS - pthreadcond NPTL fixes, new NPTL testcases Thu Feb 13 2003 Jakub Jelinek 2 3 1-45- update from CVS - include also linuxthreads FLOATINGSTACKS libs on i686 and athlon LDASSUMEKERNEL 2 2 5 to LDASSUMEKERNEL 2 4 0 is non-FLOATINGSTACKS lt, LDASSUMEKERNEL 2 4 1 to LDASSUMEKERNEL 2 4 19 is FLOATINGSTACKS lt, later is NPTL - enable TLS on alpha alphaev6- add BuildPreReq usr bin readlink Tue Feb 11 2003 Jakub Jelinek 2 3 1-44- update from CVS - pthreadonce fix Mon Feb 10 2003 Jakub Jelinek 2 3 1-43- update from CVS - vfork fix on s390- rebuilt with binutils 2 13 90 0 18-5 so that accesses to errno don t bind locally 83325 Thu Feb 06 2003 Jakub Jelinek 2 3 1-42- update from CVS - fix pthreadcreate after vfork exec in linuxthreads Wed Feb 05 2003 Jakub Jelinek 2 3 1-41- update from CVS Thu Jan 30 2003 Jakub Jelinek 2 3 1-40- update from CVS Wed Jan 29 2003 Jakub Jelinek 2 3 1-39- update from CVS - enable TLS on s390 and sparc Fri Jan 17 2003 Jakub Jelinek 2 3 1-38- update from CVS - initialize environ in glibcpostupgrade to empty array, so that it is not NULL - compat symlink for s390x enable glibc-profile on x86-64- only include on IA-32, Alpha and Sparc 32-bit Thu Jan 16 2003 Jakub Jelinek 2 3 1-37- update from CVS - nscd fixes, scanf fix - fix nptlarches noarch build 81909 - IA-64 TLS fixes Tue Jan 14 2003 Jakub Jelinek 2 3 1-36- update from CVS - rework - debuginfo subpackage, add - debuginfo-common subpackage on IA-32, Alpha and Sparc ie auxiliary arches - fix vfork in libc a on PPC32, Alpha, Sparc - fix libio locks in linuxthreads if is dlopened later 81374 Mon Jan 13 2003 Jakub Jelinek 2 3 1-35- update from CVS - dlclose bugfixes - fix N PTL libpthread a - fix glibcpostupgrade on several arches Sat Jan 11 2003 Jakub Jelinek 2 3 1-34- update from CVS - TLS support on IA-64 Wed Jan 08 2003 Jakub Jelinek 2 3 1-33- fix vfork in linuxthreads 81377, 81363 Tue Jan 07 2003 Jakub Jelinek 2 3 1-32- update from CVS - don t use TLS libs if kernel doesn t set ATSYSINFO 80921, 81212 - add ntpadjtime on alpha 79996 - fix nptldb 81116 Sun Jan 05 2003 Jakub Jelinek 2 3 1-31- update from CVS - support all architectures again Fri Jan 03 2003 Jakub Jelinek 2 3 1-30- fix condvar compatibility wrappers - add ugly hack to use non-TLS libs if a binary is seen to have errno, herrno or res symbols in Fri Jan 03 2003 Jakub Jelinek 2 3 1-29- update from CVS - fixes for new condvar Thu Jan 02 2003 Jakub Jelinek 2 3 1-28- new NPTL condvar implementation plus related linuxthreads symbol versioning updates Thu Jan 02 2003 Jakub Jelinek 2 3 1-27- update from CVS - fix include with - DBSDSOURCE or without feature set macros - make sigaction, sigwait and raise the same between - lpthread - lc and - lc - lpthread in linuxthreads builds Tue Dec 31 2002 Jakub Jelinek 2 3 1-26- fix dlclose Sun Dec 29 2002 Jakub Jelinek 2 3 1-25- enable sysenter by default for now - fix endless loop in ldconfig Sat Dec 28 2002 Jakub Jelinek 2 3 1-24- update from CVS Fri Dec 27 2002 Jakub Jelinek 2 3 1-23- update from CVS - fix ptmallocinit after clearenv 80370 Sun Dec 22 2002 Jakub Jelinek 2 3 1-22- update from CVS - add IA-64 back - move TLS libraries from lib i686 to lib tls Thu Dec 19 2002 Jakub Jelinek 2 3 1-21- system 3 fix for linuxthreads - don t segfault in pthreadattrinit from add cancellation tests from nptl to linuxthreads Wed Dec 18 2002 Jakub Jelinek 2 3 1-20- fix up lists of exported symbols their versions from the libraries Wed Dec 18 2002 Jakub Jelinek 2 3 1-19- fix --with-tls --enable-kernel 2 2 5 libc on IA-32 Wed Dec 18 2002 Jakub Jelinek 2 3 1-18- update from CVS - fix NPTL hanging mozilla - initialize malloc in mALLOPt fixes problems with squid, 7995 7 - make linuxthreads work with dldynamicweak 0 - clear dldynamicweak everywhere Tue Dec 17 2002 Jakub Jelinek 2 3 1-17- update from CVS - NPTL socket fixes, flockfile ftrylockfile funlockfile fix - kill - debug sub-package, rename - debug-static to - debug - clear dldynamicweak for NPTL Mon Dec 16 2002 Jakub Jelinek 2 3 1-16- fix and for C - automatically generate NPTL libpthread wrappers Mon Dec 16 2002 Jakub Jelinek 2 3 1-15- update from CVS - all functions which need cancellation should now be cancellable both in and - removed AATTAATTGLIBC2 3 2 cancellation wrappers Fri Dec 13 2002 Jakub Jelinek 2 3 1-14- update from CVS - replace libclockneededAATTGOTOFF ebx with gs offsetof tcbheadt, multiplethreads - start of new NPTL cancellation wrappers Thu Dec 12 2002 Jakub Jelinek 2 3 1-13- update from CVS - use inline locks in malloc Tue Dec 10 2002 Jakub Jelinek 2 3 1-12- update from CVS - support LDASSUMEKERNEL 2 2 5 in statically linked programs Mon Dec 09 2002 Jakub Jelinek 2 3 1-11- upda te from CVS - rebuilt with gcc-3 2 1-2 Fri Dec 06 2002 Jakub Jelinek 2 3 1-10- update from CVS - non-nptl --with-tls --without-thread FLOATINGSTACKS libpthread should work now - faster libc locking when using nptl - add OUTPUTFORMAT to linker scripts - fix x8664 sendfile 79111 Wed Dec 04 2002 Jakub Jelinek 2 3 1-9- update from CVS - RUSCII support 78906 - for nptl builds add BuildRequires - fix byteswap h for non-gcc 77689 - add nptl-devel package Tue Dec 03 2002 Jakub Jelinek 2 3 1-8- update from CVS - make --enable-kernel 2 2 5 --with-tls --without-thread load nptl and other --with-thread libs - disable nptl by default for now Wed Nov 27 2002 Jakub Jelinek 2 3 1-7- update from CVS - restructured redhat Makefile and spec, so that contains glibc - glibc-redhat - and added nptl Fri Nov 08 2002 Jakub Jelinek 2 3 1-6- update from CVS - even more regex fixes - run sed testsuite to check glibc regex Thu Oct 24 2002 Jakub Jelinek 2 3 1-5- fix LDDEBUG statistics and LDTRACEPRELINKING in programs using T hu Oct 24 2002 Jakub Jelinek 2 3 1-4- update from CVS - fixed a and A in printf 75821 - fix recomp memory leaking 76594 Tue Oct 22 2002 Jakub Jelinek 2 3 1-3- update from CVS - some more regex fixes - fix libpthread a 76484 - fix locale-archive enlarging Fri Oct 18 2002 Jakub Jelinek 2 3 1-2- update from CVS - don t need to use 128K of stacks for DNS lookups - regex fixes - updated timezone data e g for this year s Brasil DST changes - expand in RPATH RUNPATH dlopen filenames Fri Oct 11 2002 Jakub Jelinek 2 3 1-1- update to 2 3 1 final - support really low thread stack sizes 74073 - tzdata update Wed Oct 09 2002 Jakub Jelinek 2 3-2- update from CVS - handle low stack limits - move s390x into lib64 Thu Oct 03 2002 Jakub Jelinek 2 3-1- update to 2 3 final - fix freopen on libstdc 2 2 94-3- don t prelink - r on ppc x86-64 sparc , it doesn t speed things up, because they are neither REL arches, nor ELFMACHINERELRELATIVE - fix sparc64 build Sun Sep 29 2002 Jakub Jelinek 2 2 94-2- update from C VS Sat Sep 28 2002 Jakub Jelinek 2 2 94-1- update from CVS - prelink on ppc and x86-64 too - don t remove ppc memset - instead of listing on which arches to remove glibc-compat list where it should stay Fri Sep 06 2002 Jakub Jelinek 2 2 93-5- fix wcsmbs functions with invalid character sets or malloc failures - make sure ctypeb etc compat vars are updated even if they are copy relocs in the main program Thu Sep 05 2002 Jakub Jelinek 2 2 93-4- fix missing setherrno definition leading to unresolved setherrno symbol Wed Sep 04 2002 Jakub Jelinek 2 2 93-3- security fix - increase dns-network c MAXPACKET to at least 65536 to avoid buffer overrun Likewise glibc-compat dns - c Tue Sep 03 2002 Jakub Jelinek 2 2 93-2- temporarily add back ctypeb, ctypetolower and ctypetoupper to libc a and export them as AATTAATTGLIBC2 0 symbols, not AATTGLIBC2 0 from - we have still lots of a libraries referencing ctype out there Tue Sep 03 2002 Jakub Jelinek 2 2 93-1- update from CVS - 2 2 93 release - use double instead of single indirection in isXXX macros - per-locale wcsmbs conversion state Sat Aug 31 2002 Jakub Jelinek 2 2 92-2- update from CVS - fix newlocale duplocale uselocale - disable profile on x8664 for now Sat Aug 31 2002 Jakub Jelinek 2 2 92-1- update from CVS - 2 2 92 release - fix gettext after uselocale - fix locales in statically linked threaded programs - fix NSS Thu Aug 29 2002 Jakub Jelinek 2 2 91-1- update from CVS - 2 2 91 release - fix fd leaks in locale-archive reader 72043 - handle EROFS in build-locale-archive gracefully 71665 Wed Aug 28 2002 Jakub Jelinek 2 2 90-27- update from CVS - fix rematch 72312 - support more than 1024 threads Fri Aug 23 2002 Jakub Jelinek 2 2 90-26- update from CVS - fix i386 build Thu Aug 22 2002 Jakub Jelinek 2 2 90-25- update from CVS - fix locale-archive loading hang on some non-primary locales 72122, 71878 - fix umount problems with locale-archives when usr is a separate partition 72043 - add LICENSES file Fri Aug 16 2002 Jakub Jelinek 2 2 90-24- update from CVS - only mmap up to 2MB of locale-archive on 32-bit machines initially - fix fseek past end fread segfault with mmaped stdio - include which is mistakenly not included in glibc-devel on IA-32 Fri Aug 16 2002 Jakub Jelinek 2 2 90-23- don t return normalized locale name in setlocale when using locale-archive Thu Aug 15 2002 Jakub Jelinek 2 2 90-22- update from CVS - optimize for primary system locale - localedef fixes 71552, 67705 Wed Aug 14 2002 Jakub Jelinek 2 2 90-21- fix path to locale-archive in libc reader - build locale archive at glibc-common post time - export strtoldinternal and wcstoldinternal on Alpha again - workaround some localedata problems Tue Aug 13 2002 Jakub Jelinek 2 2 90-20- update from CVS - patch out setthreadarea for now Fri Aug 09 2002 Jakub Jelinek 2 2 90-19- update from CVS - GB18030 patch from Yu Shao - applied Debian patch for getaddrinfo IPv4 vs IPv6- fix regcomp 71039 Sun Aug 04 2002 Jakub Jelinek 2 2 90-18- update from CVS - use usr sbin pr elink, not prelink 70376 Thu Jul 25 2002 Jakub Jelinek 2 2 90-17- update from CVS Thu Jul 25 2002 Jakub Jelinek 2 2 90-16- update from CVS - ungetc fix 69586 - fseek errno fix 69589 - change etrlimit prototypes for C 68588 - use --without-tls instead of --disable-tls Thu Jul 11 2002 Jakub Jelinek 2 2 90-15- set nscd user s shell to sbin nologin 68369 - fix glibc-compat buffer overflows security - buildrequire prelink, don t build glibc s own copy of it 67567 - update from CVS - regex fix 67734 - fix unused warnings 67706 - fix freopen with mmap stdio 67552 - fix realloc 68499 Tue Jun 25 2002 Bill Nottingham 2 2 90-14- update from CVS - fix argp on long words - update atime in libio Sat Jun 22 2002 Jakub Jelinek 2 2 90-13- update from CVS - a thread race fix - fix readdir on invalid dirp Wed Jun 19 2002 Jakub Jelinek 2 2 90-12- update from CVS - don t use thread in headers - fix system 3 in threaded apps - update prelink, so that it is possible to prelink - u on Alpha Fri Jun 07 2002 Jakub Jelinek 2 2 90-11- update from CVS - fix moddi3 65612, 65695 - fix etherline 64427 - fix setvbuf with mmap stdio 65864 - --disable-tls for now, waiting for kernel - avoid duplication of divtf3 etc on IA-64- make sure get entr and IOwfilejumps are exported 62278 Tue May 21 2002 Jakub Jelinek 2 2 90-10- update from CVS - fix Alpha pthread bug with gcc 3 1 Fri Apr 19 2002 Jakub Jelinek 2 2 5-35- fix nice Mon Apr 15 2002 Jakub Jelinek 2 2 5-34- add relocation dependencies even for weak symbols 63422 - stricter checkfds check for suid sgid binaries - run make check at install time Sat Apr 13 2002 Jakub Jelinek 2 2 5-33- handle Dec 31 1969 in mktime for timezones west of GMT 63369 - back out do-lookup h change 63261, 63305 - use memory clobber instead all the fancy stuff in i386 i686 bits string h since lots of compilers break on it - fix sparc build with gcc 3 1- fix spec file for athlon Tue Apr 09 2002 Jakub Jelinek 2 2 5-32- fix debugging of threaded apps 62804 - fix DST for Estonia 61494 - document that pthreadmutexattr etkindnp are deprecated and pthreadmutexattr ettype should be used instead in man pages 61485 - fix undefined externals Fri Apr 05 2002 Jakub Jelinek 2 2 5-31- temporarily disable prelinking as some statically linked binaries linked against debugging versions of old glibcs die on it 62352 - fix for - std c99 62516 - fix etherntohost segfault 62397 - remove in glibcpostupgrade on i386 all lib i686 libc - lib i686 libm - and lib i686 libpthread - not just current version 61633 - prelink - r on alpha too Thu Mar 28 2002 Jakub Jelinek 2 2 5-30- update GB18030 iconv module Yu Shao Tue Mar 26 2002 Jakub Jelinek 2 2 5-29- features h fix Tue Mar 26 2002 Jakub Jelinek 2 2 5-28- update from CVS - fix nscd with huge groups - fix nis to not close fds it shouldn t - rebuilt against newer glibc-kernheaders to use the correct PATHMAX - handle glibc the same way as add a couple of locales 61922 - readd temporarily currencies which were superceeded by Euro into the list of acce pted currencies by localedef to make standard conformance testsuites happy - temporarily moved libcwaitpid back to make Sun JDK happy - use old malloc code - prelink i686 athlon and prelink - r i686 athlon Thu Mar 14 2002 Jakub Jelinek 2 2 5-27- update from CVS - fix DST handling for southern hemisphere 60747 - fix daylight setting for tzset 59951 - fix ftime 60350 - fix nice return value - fix a malloc segfault - temporarily moved libcwait, libcfork and libcstackend back to what they used to be exported at - censorship 60758 Thu Feb 28 2002 Jakub Jelinek 2 2 5-26- update from CVS - use attribute visibility if supported, use rtldlocal for only objects - provide libc s own di3 Wed Feb 27 2002 Jakub Jelinek 2 2 5-25- switch back to 2 2 5, mmap stdio needs work Mon Feb 25 2002 Jakub Jelinek 2 2 90-8- fix two other mmap stdio bugs 60228 Thu Feb 21 2002 Jakub Jelinek 2 2 90-7- fix yet another mmap stdio bug 60145 Tue Feb 19 2002 Jakub Jelinek 2 2 90-6- fix mmap stdio bug seen on ld as File truncat ed error, 60043 - apply Andreas Schwab s fix for pthread sigwait - remove lib i686 libraries in glibcpostupgrade when performing i386 glibc install Thu Feb 14 2002 Jakub Jelinek 2 2 90-5- update to CVS - added glibc-utils subpackage - disable autoreq in glibc-debug - readd lang to locale files Thu Feb 07 2002 Jakub Jelinek 2 2 90-4- update to CVS - move glibc private symbols to GLIBCPRIVATE symbol version Wed Jan 09 2002 Jakub Jelinek 2 2 90-3- fix a sqrt bug on alpha which caused SHNUNDEF symbol in libm Tue Jan 08 2002 Jakub Jelinek 2 2 90-2- add debug-static package Mon Dec 31 2001 Jakub Jelinek 2 2 90-1- update from CVS - remove - DUSESTRINGINLINES - add debug subpackage to trim glibc and glibc-devel size Wed Oct 03 2001 Jakub Jelinek 2 2 4-19- fix strsep Fri Sep 28 2001 Jakub Jelinek 2 2 4-18- fix a bug with duplicate searchlists in lscope - fix erfcl - inf - turn into linker script Wed Sep 26 2001 Jakub Jelinek 2 2 4-17- fix a lookup bug after lots of dlopen calls - fix CMSGDATA for non-gcc non-ISOC99 compilers 53984 - prelinking support for Sparc64 Fri Sep 21 2001 Jakub Jelinek 2 2 4-16- update from CVS to fix DTSYMBOLIC - prelinking support for Alpha and Sparc Tue Sep 18 2001 Jakub Jelinek 2 2 4-15- update from CVS - linuxthreads now retries if -1 EINTR is returned from reading or writing to thread manager pipe 43742 - use DTFILTER in 53394 - update glibc prelink patch so that it handles filters - fix timer with SIGEVNONE 53494 - make glibcpostupgrade work on PPC patch from Franz Sirl Mon Sep 10 2001 Jakub Jelinek 2 2 4-14- fix build on sparc32- 2 2 4-13 build for some reason missed some locales on alpha ia64 Mon Sep 03 2001 Jakub Jelinek 2 2 4-13- fix iconvconfig Mon Sep 03 2001 Jakub Jelinek 2 2 4-12- add fam to etc rpc 52863 - fix for C 52960 - fix perror Mon Aug 27 2001 Jakub Jelinek 2 2 4-11- fix strnlen x, -1 Mon Aug 27 2001 Jakub Jelinek 2 2 4-10- doh, should only define libcrwlockt if USEUNIX98 Mon Aug 27 2001 Jakub Jelinek 2 2 4-9- fix bits libc-lock h so that gc c can compile - fix s390 build Fri Aug 24 2001 Jakub Jelinek 2 2 4-8- kill stale library symlinks in ldconfig 52350 - fix inttypes h for G 2 2 4-7- fix strnlen on IA-64 50077 Thu Aug 16 2001 Jakub Jelinek 2 2 4-6- glibc 2 2 4 final - fix - lpthread - static 51672 Fri Aug 10 2001 Jakub Jelinek 2 2 4-5- doh, include libio tst-swscanf c Fri Aug 10 2001 Jakub Jelinek 2 2 4-4- don t crash on catclose -1 - fix wscanf handling - fix return value from swprintf - handle year U W week week day in strptime Thu Aug 09 2001 Jakub Jelinek 2 2 4-3- update from CVS to - fix strcoll 50548 - fix seekdir 51132 - fix memusage 50606 - don t make config file, just don t verify its content Mon Aug 06 2001 Jakub Jelinek - fix strtod and scanf 50723, 50724 Sat Aug 04 2001 Jakub Jelinek - update from CVS - fix iconv cache handling - glibc should not own , nor man3 50673 - add as emtpy config file 50699 - only run iconvconfig if usr is mounted read-write 50667 Wed Jul 25 2001 Jakub Jelinek - move iconvconfig from glibc - common into glibc subpackage, call it from glibcpostupgrade instead of common s post Tue Jul 24 2001 Jakub Jelinek - turn off debugging printouts in iconvconfig Tue Jul 24 2001 Jakub Jelinek - update from CVS - fix IA-32 makecontext - make fflush 0 thread-safe 46446 Mon Jul 23 2001 Jakub Jelinek - adjust prelinking DT and SHT values in elf h - update from CVS - iconv cache - make iconv work in SUID SGID programs 34611 Fri Jul 20 2001 Jakub Jelinek - update from CVS - kill non-pic code in - fix getdate - fix some locales 49402 - rebuilt with binutils-2 11 90 0 8-5 to place section properly in and add floating stacks on IA-64, Alpha, Sparc 49308 Mon Jul 16 2001 Jakub Jelinek - make lib i686 directory owned by glibc Mon Jul 09 2001 Jakub Jelinek - remove rquota hx headers which are now provided by quota 47141 - add prelinking patch Thu Jul 05 2001 Jakub Jelinek - require sh-utils for nscd Mon Jun 25 2001 Jakub Jelinek - update from CVS 43681, 43350, 44663, 45685 - fix roRO bug 44644 Wed J un 06 2001 Jakub Jelinek - fix a bunch of math bugs 43210, 43345, 43346, 43347, 43348, 43355 - make rpc headers - ansi compilable 42390 - remove alphaev6 optimized memcpy, since there are still far too many broken apps which call memcpy where they should call memmove - update from CVS to among other things - fix tanhl bug 43352 Tue May 22 2001 Jakub Jelinek - fix include with - DXOPENSOURCE 500 on ia64 35968 - fix a dlclose reldeps handling bug - some more profiling fixes - fix tgmath h Thu May 17 2001 Jakub Jelinek - make ldconfig more quiet - fix LDPROFILE on i686 41030 Wed May 16 2001 Jakub Jelinek - fix the hardlink program, so that it really catches all files with identical content - add a s390x clone fix Wed May 16 2001 Jakub Jelinek - fix rpc for non-threaded apps using svcfdset and similar variables 40409 - fix nss compatibility DSO versions for alphaev6- add a hardlink program instead of the shell 3x for plus cmp - s link which takes a lot of time during build - rework BuildPreReq and Conflicts with gcc, so that it applies only where it has to Fri May 11 2001 Jakub Jelinek - fix locale name of jaJP in UTF-8 39783 - fix research2 40244 - fix memusage script 39138, 39823 - fix dlsym RTLDNEXT, from main program 39803 - fix xtrace script 39609 - make glibc conflict with glibc-devel 2 2 2 and below to make sure libcnonshared a has atexit - fix getconf LFSCFLAGS on 64bitters - recompile with gcc-2 96-84 or above to fix binary compatibility problem with framestatefor function 37933 Fri Apr 27 2001 Jakub Jelinek - glibc 2 2 3 release - fix strcoll 36539 - add BuildPreReqs 36378 Wed Apr 25 2001 Jakub Jelinek - update from CVS Fri Apr 20 2001 Jakub Jelinek - update from CVS - fix sparc64, ia64 - fix some locale syntax errors 35982 Wed Apr 18 2001 Jakub Jelinek - update from CVS Wed Apr 11 2001 Jakub Jelinek - update from CVS Fri Apr 06 2001 Jakub Jelinek - support even 2 4 0 kernels on ia64, sparc64 and s390x - include UTF-8 locales - make gconv-modules config noreplace Fri Mar 23 2001 Jakub Jelinek - back out sunrpc changes Wed Mar 21 2001 Jakub Jelinek - update from CVS - fix ia64 build - fix pthreadgetattrnp Fri Mar 16 2001 Jakub Jelinek - update from CVS - run atexit registered functions at dlclose time if they are in shared libraries 28625 - add pthreadgetattrnp API to make JVM folks happy Wed Mar 14 2001 Jakub Jelinek - require 2 4 1 instead of 2 4 0 on platforms where it required 2 4 kernel - fix ldd behaviour on unresolved symbols - remove nonsensical ldconfig warning, update osversion for the most recent library with the same soname in the same directory instead 31703 - apply selected patches from CVS - s390x spec file changes from Florian La Roche Wed Mar 07 2001 Jakub Jelinek - fix gencat 30894 - fix ldconfig changes from yesterday, fix LDASSUMEKERNEL handling Tue Mar 06 2001 Jakub Jelinek - update from CVS - make pthreadattrsetstacksize consistent before and after pthread manager is started 28194 - pass back struct sigcontext from pthread signal wrapp er on ia32 only so far, 28493 - on i686 ship both --enable-kernel 2 2 5 and 2 4 0 libc libm libpthread, make pick the right one Sat Feb 17 2001 Preston Brown.- glib-common doesn t require glibc, until we can figure out how to get out of dependency hell Sat Feb 17 2001 Jakub Jelinek - make glibc require particular version of glibc-common and glibc-common prerequire glibc Fri Feb 16 2001 Jakub Jelinek - glibc 2 2 2 release - fix regex REGICASE bug seen in ksymoops Sat Feb 10 2001 Jakub Jelinek - fix regexec leaking memory 26864 Fri Feb 09 2001 Jakub Jelinek - update from CVS - fix ia64 build with gnupro - make regex 64bit clean - fix tgmath make check failures on alpha Tue Feb 06 2001 Jakub Jelinek - update again for ia64 DF1INITFIRST Fri Feb 02 2001 Jakub Jelinek - update from CVS - fix getaddrinfo 25437 - support DF1INITFIRST 25029 Wed Jan 24 2001 Jakub Jelinek - build all auxiliary arches with --enablekernel 2 4 0, those wanting to run 2 2 kernels can downgrade to the base architectur e glibc Sat Jan 20 2001 Jakub Jelinek - remove lang flags from lib locale files temporarily Sun Jan 14 2001 Jakub Jelinek - update to 2 2 1 final - fix a pthreadkillotherthreadsnp breakage 23966 - make static binaries using dlopen work on ia64 again - fix a typo in glibc-common group Wed Jan 10 2001 Bernhard Rosenkraenzer - devel requires glibc - noreplace Wed Jan 10 2001 Jakub Jelinek - some more security fixes - don t look up LDPRELOAD libs in cache for SUID apps because that bypasses SUID bit checking on the library - place output files for profiling SUID apps into var profile, use ONOFOLLOW for them - add checks for MEMUSAGEOUTPUT and SEGFAULTOUTPUTNAME - hardlink identical locale files together - add lang tags to locale stuff - remove for now, it is provided by locale-utf8 package Mon Jan 08 2001 Jakub Jelinek - add glibc-common subpackage - fix alphaev6 memcpy 22494 - fix sys cdefs h 22908 - don t define stdin stdout stderr as macros for - traditional 22913 - work around a bug in IBM J DK 22932, 23012 - fix pmapunset when network is down 23176 - move nscd in rc d before netfs on shutdown - fix RESOLVHOSTCONF in SUID apps 23562 Fri Dec 15 2000 Jakub Jelinek - fix ftw and nftw Wed Dec 13 2000 Jakub Jelinek - fix fcvt 22184 - ldd is not crashing any longer again 22197 - fix gencat Mon Dec 11 2000 Jakub Jelinek - fix alpha htonl and alphaev6 stpcpy Sat Dec 09 2000 Jakub Jelinek - update to CVS to - fix getnameinfo 21934 - don t stomp on memory in rpath handling 21544 - fix setlocale 21507 - fix loading code 21579 - use auxarches define in spec file for auxiliary architectures 21219 - remove usr share directory from filelist 21218 Sun Nov 19 2000 Jakub Jelinek - update to CVS to fix getaddrinfo Fri Nov 17 2000 Jakub Jelinek - update to CVS to fix freopen - remove all alpha workarounds, not needed anymore Wed Nov 15 2000 Jakub Jelinek - fix dladdr bug on alpha sparc32 sparc64- fix Makefiles so that they run static tests properly Tue Nov 14 2000 Jakub Jelinek - update to CVS to fix ldconfig Thu Nov 09 2000 Jakub Jelinek - update to glibc 2 2 release Mon Nov 06 2000 Jakub Jelinek - update to CVS to - export sysconfAATTAATTGLIBC2 2 20417 Fri Nov 03 2000 Jakub Jelinek - merge to 2 1 97 Mon Oct 30 2000 Jakub Jelinek - update to CVS, including - fix WORDBIT LONGBIT definition in limits h 19088 - fix hesiod 19375 - set LCMESSAGES in zic zdump for proper error message output 19495 - fix LFS fcntl when used with non-LFS aware kernels 19730 Thu Oct 19 2000 Jakub Jelinek - fix alpha semctl 19199 - update to CVS, including - fix glibc headers for Compaq non-gcc compilers - fix locale alias handling code 18832 - fix rexec on little endian machines 18886 - started writing changelog again Thu Aug 10 2000 Adrian Havill - added ja ujis alias for backwards compatibility. Neither I Nor II. question 19 name 7.Geneticists say that they are very close to confirming the genetic roots of psychiatric illnesses such as depression and schizophrenia, and consequently, that doctors will be able to eradicate these diseases through early identification and gene therapy. On which of the following assumptions does the statement above rely. Strategies are now available for eliminating psychiatric illnesses. Certain psychiatric illnesses have a genetic basis. All human diseases can be traced back to genes and how they are expressed. In the future, genetics will become the only relevant field for identifying psychiatric illnesses. question 29 name 7.Consider the following Boolean expression for F.