How important is it to not user gcc2.95.2 for kernel 2.4?
The docs for kernel-2.4.4 say to not use 2.95.2 but use 2.91.66 instead? If I must use the older compiler, can I install this along side 2.95.2 and if so, how do I build using it instead? Thanks, CC
* Just Another SuSE User
The docs for kernel-2.4.4 say to not use 2.95.2 but use 2.91.66 instead? If I must use the older compiler, can I install this along side 2.95.2 and if so, how do I build using it instead?
You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas. IIRC the good Mr. Thomas provided update rpm's for us :-) -- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas.
IIRC the good Mr. Thomas provided update rpm's for us :-)
Well That's Good , but is it any better than 2.95.2 ?? Is there any more info - or should we trust by blind faith, that it will not cause problems that 2.95.2 may. What parts of compiling a kernel could there be some affect ? Why is ergs circa 1999 , the prefered choice in the /usr/src/linux/documentation/changes ? egcs 1.1.2 (gcc 2.91.66) - ftp://sourceware.cygnus.com/pub/gcc/releases/egcs-1.1.2/egcs-1.1.2.tar.bz2 I can probably answer this myself - because it was a damn good compiler. But who would have known that it could be used to compile the latest kernels .? especially with all the 'updating hype' going on reminds me of -m$ I know don't say it . -semi-rant- Else I would not have updated to gcc2.95.2 -which is not as good as egcs 1.1.2 . By the time I find this out , I have already compiled kernels 4 times using 2.95.2 I have had quite dubious results. see on this list - the many, ''problems compiling . . .'' messages posted - march-april - . Dubious results may or may not have been caused by the compiler though - but I wish I knew more of what is at fault so I can get on with making my cottage cheese : \ So lets talk compilers once in a while here on the list . -end-semi-rant-
Quote from ''Just Another SuSE User'' The docs for kernel-2.4.4 say to not use 2.95.2 but use 2.91.66 instead? If I must use the older compiler, can I install this along side 2.95.2 and if so, how do I build using it instead?
Well I have an egcs 1.1.2 (gcc 2.91.66) untared and ready to use /usr/src/egcs 1.1.2 . I guess the best way is to try to compile with it , and decide for myself that it suites my needs. I can always re-install gcc2.95.3rpm . You could install the available 2.95.3rpm from ftp.suse.com/pub/people/pthomas and just compile as usual to try it out . Looks like you shouldn't be using 2.95.2 any more now that we know there could be problems with it. John W Mislan On Wed, 02 May 2001, Mads Martin Jørgensen wrote:
* Just Another SuSE User
[May 02. 2001 12:23]: The docs for kernel-2.4.4 say to not use 2.95.2 but use 2.91.66 instead? If I must use the older compiler, can I install this along side 2.95.2 and if so, how do I build using it instead?
You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas. IIRC the good Mr. Thomas provided update rpm's for us :-)
-- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com --
God does not play dice" -- Einstein "Not only does God play dice, he sometimes throws them where they can't be seen." -- Stephen Hawking
* basslake
You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas.
IIRC the good Mr. Thomas provided update rpm's for us :-)
Well That's Good , but is it any better than 2.95.2 ?? Is there any more info - or should we trust by blind faith, that it will not cause problems that 2.95.2 may.
Well, since it is a minor release, it is only bugfixes because 2.95.2 was so buggy. Go look at gcc.gnu.org, and you'll see.
What parts of compiling a kernel could there be some affect ?
Why is ergs circa 1999 , the prefered choice in the /usr/src/linux/documentation/changes ? egcs 1.1.2 (gcc 2.91.66) - ftp://sourceware.cygnus.com/pub/gcc/releases/egcs-1.1.2/egcs-1.1.2.tar.bz2 I can probably answer this myself - because it was a damn good compiler. But who would have known that it could be used to compile the latest kernels .? especially with all the 'updating hype' going on reminds me of -m$ I know don't say it . -semi-rant- Else I would not have updated to gcc2.95.2 -which is not as good as egcs 1.1.2 .
By the time I find this out , I have already compiled kernels 4 times using 2.95.2 I have had quite dubious results. see on this list - the many, ''problems compiling . . .'' messages posted - march-april - . Dubious results may or may not have been caused by the compiler though - but I wish I knew more of what is at fault so I can get on with making my cottage cheese : \ So lets talk compilers once in a while here on the list .
-end-semi-rant-
egcs-1.1.2 is mentioned because the documentation is out of date. -- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
Thanks Mads I will take your advice to look at gcc.gnu.org.
egcs-1.1.2 is mentioned because the documentation is out of date.
This is really really bad that the doc was able to tell that we needed specified requirements for the 2.4.2 update - and then point us to versions of those requirements that may be in error. I wonder though, how egcs-1.1.2 would handle compiling present kernels. not to contradict you , but other parts of the doc were correct - the modutil - util-linux - ppp - , and most else that I see were correct . Quote from doc /usr/src/linux/documentation/changes on gcc compilers 2.95 - series Similarly, if you choose to use gcc-2.95 or derivatives, be sure not to use -fstrict-aliasing (which, depending on your version of gcc 2.95, may necessitate using -fno-strict-aliasing). I have been seeing this all the time in compiles '' -fstrict-aliasing '' the doc says not to have this option . I wonder why I'm seeing it in compiles when I didn't do , make -fstrict-aliasing .? ''your version of gcc 2.95, may necessitate using -fno-strict-aliasing).'' I wonder then also do I need to add this option to compile kernels -fno-strict-aliasing.? This option makes the compiled, result in more compact code (smaller size) ? I believe. that's a good thing that I would like to be able to use. but that doc looks pretty up to date to me now that I go back and check. except for the egcs part , but I bet that egcs will work to compile new kernels. maybe those kernel guys know some things that the rest of us don't know. like for instance that egcs 1.1.2 (gcc 2.91.66) may work great to compile new kernels. Damn I'm iching to give it a try . egcs always ran so smooth . I'll probably give a try to the gcc 2.95.3 ftp.suse.com/pub/people/pthomas first though. bye for now John On Wed, 02 May 2001, Mads Martin Jørgensen wrote:
* basslake
[May 02. 2001 14:03]: You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas.
IIRC the good Mr. Thomas provided update rpm's for us :-)
Well That's Good , but is it any better than 2.95.2 ?? Is there any more info - or should we trust by blind faith, that it will not cause problems that 2.95.2 may.
Well, since it is a minor release, it is only bugfixes because 2.95.2 was so buggy. Go look at gcc.gnu.org, and you'll see.
What parts of compiling a kernel could there be some affect ?
Why is ergs circa 1999 , the prefered choice in the /usr/src/linux/documentation/changes ? egcs 1.1.2 (gcc 2.91.66) - ftp://sourceware.cygnus.com/pub/gcc/releases/egcs-1.1.2/egcs-1.1.2.tar.bz2 I can probably answer this myself - because it was a damn good compiler. But who would have known that it could be used to compile the latest kernels .? especially with all the 'updating hype' going on reminds me of -m$ I know don't say it . -semi-rant- Else I would not have updated to gcc2.95.2 -which is not as good as egcs 1.1.2 .
By the time I find this out , I have already compiled kernels 4 times using 2.95.2 I have had quite dubious results. see on this list - the many, ''problems compiling . . .'' messages posted - march-april - . Dubious results may or may not have been caused by the compiler though - but I wish I knew more of what is at fault so I can get on with making my cottage cheese : \ So lets talk compilers once in a while here on the list .
-end-semi-rant-
egcs-1.1.2 is mentioned because the documentation is out of date.
-- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com --
God does not play dice" -- Einstein "Not only does God play dice, he sometimes throws them where they can't be seen." -- Stephen Hawking
John, no offence meant, but please take a look at http://learn.to/quote. Quoting selectively and answering after the quote makes mails *much* more readable. * basslake [Wed, 2 May 2001 17:27:54 -0400]:
I wonder though, how egcs-1.1.2 would handle compiling present kernels.
It will compile them, no question. That part of the documentation is way out of date. All the kernels for 7.2 are compiled with gcc 2.95.3 with no problems whatsoever.
I have been seeing this all the time in compiles '' -fstrict-aliasing '' the doc says not to have this option .
AFAIR, the publicly released gcc 2.95.0 had strict aliasing turned on. This led to miscompiled kernels because the kernel source code is violating the ANSI C standard in a few places, most notably in the network stack. A detailed description would require intimate knowledge of C, which I can't presume. Because of the problems it was switched off in later versions. BTW, in most of the cases where the compiler miscompiled the kernel it was the kernel source that did something wrong, like incorrectly written inline assembly code.
I wonder then also do I need to add this option to compile kernels -fno-strict-aliasing.?
The Makefile automatically adds -fno-strict-aliasing to the compiler flags.
This option makes the compiled, result in more compact code (smaller size) ? I believe. that's a good thing that I would like to be able to use.
Not necessarily, but it allows the compiler do some optimisations it otherwise can't do. As the kernel developers won't change the source because they don't like the correct way to do it, there is no alternative to -fno-strict-aliasing. BTW, the upcoming GCC 3.0 will most probably have strict aliasing turned on The rationale is that it will force people to write correct code. The downside is, that it'll probably force distributors like us to do a lot of bug hunting.
I'll probably give a try to the gcc 2.95.3 ftp.suse.com/pub/people/pthomas first though.
Do it, I'll assure you it works smoothly ;-) And you'll also get an improved C++ compiler with quite a few bugs removed. Philipp -- customer: My computer is running Windows98. hotline : And? customer: The machine doesn't work. hotline : You already said that.
Thanks. Just for my info, do oyu know what the problem with 2.95.2 was? -CC
* Just Another SuSE User
[May 02. 2001 12:23]: The docs for kernel-2.4.4 say to not use 2.95.2 but use 2.91.66 instead? If I must use the older compiler, can I install this along side 2.95.2 and if so, how do I build using it instead?
You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas. IIRC the good Mr. Thomas provided update rpm's for us :-)
-- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
* Just Another SuSE User
Thanks. Just for my info, do oyu know what the problem with 2.95.2 was?
2.95.2 miscompiles long long. I could for instance not get XFS working at all with 2.95.2 but with 2.95.3 it worked. -- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
Hello again I have SuSE70 - libc 2.1.3 - kernel 2.4.2-4GB gcc 2.95.2 the gcc-2.95.3 from ftp.suse.com/pub/people/pthomas/suse71 - gave me 'missing glibc' err msgs , but I want to update to the gcc 2.95.3 for my present SuSE70 setup. Are there any diffs /patches that are available for this purpose ? and if not , I have found diffs in my /usr/src/packages /sources directory but I'm not sure if they were from the gcc-2.95.3 src-rpm that I downloaded from ftp.suse.com/pub/people/pthomas. Can I just apply diffs ? which ones ? , and how ? gcc-2.95.2-complex.dif gcc-2.95.2-rs6000.diff gcc-2.95.2-s390D10-SuSE.diff gcc-2.95.2.3-s390-reload1.diff gcc-2.95.2.4-s390-2.tar.gz gcc-2.95.2.4-s390-3.tar.gz gcc-2.95.2.4-s390-4.tar.gz gcc-2.95.2.4-s390.tar.gz gcc-2.95.3-version.diff gcc-2.95.3.tar.gz could I just compile this one ? gcc-2.95.4-1.patch I'm not sure if I should try to apply some,or all of these diffs to my present gcc2.95.2 and exactly how to do it . or maybe there's an easier solution . thanks for help and info John WM On Wed, 02 May 2001, Mads Martin Jørgensen wrote:
You should use gcc 2.95.3 -- look in ftp.suse.com/pub/people/pthomas. IIRC the good Mr. Thomas provided update rpm's for us :-)
-- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogic, with just a little bit more effort." -- A. P. J.
participants (4)
-
basslake
-
Just Another SuSE User
-
Mads Martin Jørgensen
-
Philipp Thomas