On Fri, Jan 23, 2004 at 10:00:35AM -0500, Jeremy Andrews wrote:
> > Is it correct to issue "make bzImage modules modules_install"
> > or do I have to do make bzImage; make modules modules_install?
> >
> > Is there any documentation where I can read answer to this question?
>
> make help
Cool. I got to README :)
I read here "make sure you have gcc 2.95.3 available" - does it mean
my gcc-3.2.3 or gcc-3.2.2 is not suitable for kernel compiling?
Cl<
> I read here "make sure you have gcc 2.95.3 available" - does it mean
> my gcc-3.2.3 or gcc-3.2.2 is not suitable for kernel compiling?
Please have a look at http://developer.osdl.org/cherry/compile/
It should work fine.
Daniel Andersen
On Fri, Jan 23, 2004 at 05:01:23PM +0100, Daniel Andersen wrote:
> > I read here "make sure you have gcc 2.95.3 available" - does it mean
> > my gcc-3.2.3 or gcc-3.2.2 is not suitable for kernel compiling?
>
> Please have a look at http://developer.osdl.org/cherry/compile/
What if the kernel compiles cleanly but the generated code is invalid?
Or is gcc-3.2.2 BugFree(TM) (BugFree as in BugFree speech, not as
in BugFree beer)?
Cl<
On Fri, 2004-01-23 at 10:30, Karel Kulhavý wrote:
> On Fri, Jan 23, 2004 at 05:01:23PM +0100, Daniel Andersen wrote:
> > > I read here "make sure you have gcc 2.95.3 available" - does it mean
> > > my gcc-3.2.3 or gcc-3.2.2 is not suitable for kernel compiling?
> >
> > Please have a look at http://developer.osdl.org/cherry/compile/
>
> What if the kernel compiles cleanly but the generated code is invalid?
> Or is gcc-3.2.2 BugFree(TM) (BugFree as in BugFree speech, not as
> in BugFree beer)?
Many people have been using gcc-3.2 or later to build kernels, and I
haven't really heard of any problems with this, at least on i386. I
personally have used 3.2.2 and 3.3.2 (well, with Debian's patches) and
haven't had any weirdness with 2.6 or 2.4. ISTR there being arches that
need 3.x to compile, but I could be mistaken.
2.95.3 is definitely the *oldest* compiler you'd want to use, and pretty
much skip between that and 3.2.
Matt
On Fri, 2004-01-23 at 16:13, Karel Kulhavý wrote:
> On Fri, Jan 23, 2004 at 10:00:35AM -0500, Jeremy Andrews wrote:
> > > Is it correct to issue "make bzImage modules modules_install"
> > > or do I have to do make bzImage; make modules modules_install?
> > >
> > > Is there any documentation where I can read answer to this question?
> >
> > make help
>
> Cool. I got to README :)
>
> I read here "make sure you have gcc 2.95.3 available" - does it mean
> my gcc-3.2.3 or gcc-3.2.2 is not suitable for kernel compiling?
I've been compiling 2.5 and 2.6 kernels since gcc 3.3 with no problems.
In fact, there are patches on the -mm tree to help compiling with gcc
3.4 and 3.5.
I think the Documentation is a little bit updated ;-)
Matthew Reppert wrote:
>snip
>
>
>Many people have been using gcc-3.2 or later to build kernels, and I
>haven't really heard of any problems with this, at least on i386. I
>personally have used 3.2.2 and 3.3.2 (well, with Debian's patches) and
>haven't had any weirdness with 2.6 or 2.4. ISTR there being arches that
>need 3.x to compile, but I could be mistaken.
>
>2.95.3 is definitely the *oldest* compiler you'd want to use, and pretty
>much skip between that and 3.2.
>
>Matt
>
>
Same here. I've been using gcc3.2.0 and beyond currently 3.3.2 since the
day they were released and never had any big issues. I would recomend
using gcc 3.3.2 since it improves performance when using optimizations
quite a bit as far as I can remember the statistics.
Stef
On Sat, Jan 24, 2004 at 12:20:02AM +0100, Stef van der Made wrote:
> Same here. I've been using gcc3.2.0 and beyond currently 3.3.2 since the
> day they were released and never had any big issues. I would recomend
> using gcc 3.3.2 since it improves performance when using optimizations
> quite a bit as far as I can remember the statistics.
On ARM at least, gcc 3.2.x seems buggy. It's along the lines of this:
3.2.0: incorrect function argument offset calculation.
3.2.x: miscompiles NEW_AUX_ENT in fs/binfmt_elf.c
(http://gcc.gnu.org/PR8896) and incorrect structure
initialisation in fs/jffs2/erase.c
I suspect that the fs/jffs2/erase.c problem is not ARM-specific, though
I'm no compiler expert.
However, gcc 3.3 seems table on ARM, and I'm not aware of any problems
with any further 3.3.x releases.
--
Russell King
Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/
maintainer of: 2.6 PCMCIA - http://pcmcia.arm.linux.org.uk/
2.6 Serial core
On Sat, 24 Jan 2004, Stef van der Made wrote:
> Matthew Reppert wrote:
> Same here. I've been using gcc3.2.0 and beyond currently 3.3.2 since the
> day they were released and never had any big issues. I would recomend
> using gcc 3.3.2 since it improves performance when using optimizations
> quite a bit as far as I can remember the statistics.
>
> Stef
Well, according to this list, gcc-3.3.2 at least has problems to compile
ALSA correctly, unless you activate framepointer support.
IB
--
"For every government X there is at least one government Y such that X
would claim that Y is a bunch of corrupt assholes. Since every government
is a bunch of corrupt assholes, every government is right at least in
one of its claims." -- Al Viro discussing politics on lkml
Ingo Buescher wrote:
>snip
>
>Well, according to this list, gcc-3.3.2 at least has problems to compile
>ALSA correctly, unless you activate framepointer support.
>
>IB
>
>
I don't seem to have any issues using ALSA since kernel 2.6.1 and gcc
3.3.2. I'm using an soundblaster live emu10k. I did have issues before
this kernel version and had to use OSS emulation. btw I'm using x86
(Athlon K7)
Cheers,
Stef
Karel Kulhav? wrote:
> Cool. I got to README :)
>
> I read here "make sure you have gcc 2.95.3 available" - does it mean
> my gcc-3.2.3 or gcc-3.2.2 is not suitable for kernel compiling?
AFAIK, the README is woefully out of date.
On Sat, 2004-01-24 at 00:48 +0000, Russell King wrote:
> I suspect that the fs/jffs2/erase.c problem is not ARM-specific, though
> I'm no compiler expert.
I think it's been seen on MIPS too.
--
dwmw2