Re: lug-bg: gcc error - Segmentation Fault (program cpp0)

[danchev@xxxxxxxxx (George Danchev)]

On Friday 15 February 2002 04:18, you wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Thursday 14 February 2002 20:45, you wrote:
> > On Thursday 14 February 2002 19:29, you wrote:
> > > Pri kompilacija na kernel ili drug paket (BIND 9.2.0 naprimer)
> > >
> > > OS: Red Hat Linux 7.2
> > > gcc ver.2.96
> > > kernel 2.4.7; 2.4.9
> > >
> > > mashinata e Pentium 150, o/c na 188, 64MB RAM.
> > >
> > > 10x predvaritelno!
> >
> > ami tokova zhivotno kato gcc-2.96 nqma officialno release-nato ot
> > developers na gcc (vizh na site-a na gcc za da se ubedish). RH sa vzeli
> > edin CVS snapshot na gcc nqkyde okolo 3.0 patch-nali sa go zdravo (s
> > eksperimantalna cel maj, neznam) i v rezultat na koeto tova gcc e zdravo
> > broken i e predvideno samo za user space kompilacii, a ne na kernels :).
> > Oba4e s RH 7.2 trqbva da vyrvi i kgcc (tova maj e istinskoto gcc2.95 , ne
> > sym mnogo siguren za versiqta no e naistina ot upstream developers na
> > gcc) . Ta s tri dumi:
> >
> > smqna na gcc s nesto po-4oveshko - 2.95.3 , 2.95.4
> > smqna na kernela - vanilla >=2.4.10  (bez 2.4.11 i 2.4.15)
> > [mislq 4e kernela kojto idva s RH 7.2 e 2.4.9-ac, smenqj go smelo s nqkoj
> > vanilla, stoto ina4e si s forked kernel - s stara VM i t.n. i po-malko
> > hora go nabludavat i pomagat pri problemi... poslednite vanilla 2.4 sa
> > dosta koravi, vypreki 4e az neiskam da se razdelqm s moq 2.4.10 zvqr ;]
>
> Hmm.
> gcc-2.96 koeto idva s RH-7.2 ne bi triabvalo da ima problemi. ( Ili pone ne
> 4ak tolkova seriozen  problem ). Po4ti edin mesec karah s nego.
> Po skoro  mi prili4a kato OOM killer-a da se "razvihria". Vypreki 4e s RH

gcc naistina se stremi da usvoqva kolkoto se mozhe pove4e pamet (za da 
priklu4i po-byrzo s kompilaciqta), no go pravi taka da se kazhe dosta 
kulturno i zakonno, ne iziskva ot kernel-a da mu zadelq vse pove4e i pove4e 
pamet do bezkraj ako tova naistina ne e vyzmozhno. Ne mislq 4e tova e 
problema. A OOM killer-a ne pribyrzva s izbivaneto na bad procesite i 
sotvetno da izduha (iz4isti) heap-a im ot pametta 4e da osvobodi zaetite ot 
tqh mempages... toj polzva dosta slozhen algoritym i si znae rabotata (koga 
da kolq i besi:)

> idva kgcc - koeto e gcc-2.91-neshto si ako si compile-vash vanila kernel
> shte se polzva gcc-to.Probvai da compilirash bez da puskash X i razni
> sharenii. Ogledai dali predi da SEGFAULT-ne gcc-to ne ti dava niakakvi
> syobshteniq za greshi v coda koito compilirash. Dori i "vanila" 2.93 ima
> edin kup bugove koito kato stigne do greshka ot koiato ne moje da se
> vyzstanvi compilaciata oshte mnogo redove programen cod SEGFAULT-va.

li4noto mi nabludenie e 4e gcc 2.95.3 i 2.95.4 sa zheleznite compilers kym 
momenta. Kakvoto im dadesh go smilat i proizvezhdat naistina stable izpylnim 
kod. a edno ot dobrite obqsneniq za gcc 2.96:
http://www.mplayerhq.hu/DOCS/faq.html
http://www.mplayerhq.hu/DOCS/gcc-2.96-3.0.html
http://www.mplayerhq.hu/DOCS/users_against_developers.html

ina4e oplakvaniq ot nego mnogo... a eto i kakvo e kazano v 
Documetation/Changes v 2.4.10 kernela v slu4aj 4e nqkoj nqma source na 2.4 
pod ruka. A preporukite za kernela sa dobra otpravna to4ka za tova kyf da e 
compiler-a (kojto da kompilira ne samo kernel razbira se). 

The recommended compiler for the kernel is gcc 2.95.3 or .4, and it
should be used when you need absolute stability. You may use gcc 3.0.x
instead if you wish, although it may cause problems. Later  versions of gcc
have not received much testing for Linux kernel compilation, and there are
almost certainly bugs (mainly, but not exclusively, in the kernel) that
will need to be fixed in order to use these compilers. In any case, using
pgcc instead of egcs or plain gcc is just asking for trouble.

Note that gcc 2.7.2.3 is no longer a supported kernel compiler. The kernel
no longer works around bugs in gcc 2.7.2.3 and, in fact, will refuse to
be compiled with it. egcs-1.1.2 has register allocation problems in very
obscure cases. We have ensured the kernel does not trip these in any known
situation. The 2.5 tree is likely to drop egcs-1.1.2 workarounds.

The Red Hat gcc 2.96 compiler subtree can also be used to build this tree.
You should ensure you use gcc-2.96-74 or later. gcc-2.96-54 will not build
the kernel correctly.

[az bih kazal zabravete za gcc-2.96, a ako go polzvate naj-dobriq izto4nik za 
otgovori otnosno problemi sa RH - tehnite mail lists i t.n., ama tova rabota 
li e ?!]

zabravih da kazha 4e na edno ot CD-tata na RH-7.2 ima i hacked versiq na 
glibc (hacknata ot RH razbira se)... edin mnogo knoledgeble tip ot tqh go e 
predlozhil i izpylnil, nari4at go object prelinking - neznam kakvo zna4i i 
kakvo sa promenili, no s nego prilozheniqta startirat adski byrzo no pyk za 
tova mogat i da brake-vat adski byrzo , vse oste e experimental oba4e... vizh 
da ne si s tova glibc - vsystnost ideq nqmam kak ste go poznaesh :) 

Vsi4ko tova go kazvam ne 4e imam nesto protiv RH, a zastoto v tehnite distros 
ima dosta iznenadi (dosta forked upstream sources, bez da ima realna pri4ina 
za tova v pove4eto slu4ai)... ponqkoga poor documented i user-a se 4udi kakvo 
i zasto stava ... report-va bugs , oba4e drugite kojto ne polzvat RH ne mogat 
da go reproducirat i syotvetno bug-a se prevrusta v mystery ... idi 4e 
razberi. Imat mail lists kydeto se obsyzhdat podobni RH specific probs.

problema mozhe i hardwaren da e - pametta, no 4oveka se e testval sigurnove4e 
s memtest86...... No  naistina edin 4ist install na RH s gcc != 2.96 i 
normalno glibc mozhe da pomogne - znaesh li kakvo e broken. Ako isksh probvaj 
da instalirash gcc-2.95 ot binary .rpm .
Ili install na nqkoe drugo _4isto_ distro za test, da se vidi kva ste e 
reakciqta. 

-- 
Greets,
fr33zb1
===========================================================================
A mail-list of Linux Users Group - Bulgaria (bulgarian linuxers)
http://www.linux-bulgaria.org/ Hosted by Internet Group Ltd. - Stara Zagora