reassign 480130 linux-libc-dev 2.6.25-1
thanks

On Thu, May 08, 2008 at 11:43:16AM +0300, Niko Tyni wrote:
> Package: perl
> Version: 5.10.0-9.1
> Severity: serious
>
> From the alpha buildd log:
>
> http://buildd.debian.org/fetch.cgi?pkg=perl;ver=5.10.0-9.1;arch=alpha;stamp=1210220263
>
> `sh cflags "optimize='-O2 -g'" perl.o` perl.c
> CCCMD = cc -DPERL_CORE -c -D_REENTRANT -D_GNU_SOURCE -DDEBIAN -fno-strict-aliasing -pipe -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -std=c89 -O2 -g -Wall -ansi -pedantic -W -Wextra -Wdeclaration-after-statement -Wendif-labels -Wc++-compat -DPERL_GCC_PEDANTIC
> perl.c: In function 'perl_construct':
> perl.c:388: error: 'CONFIG_HZ' undeclared (first use in this function)

> 5.10.0-9 built successfully on the same buildd host, so changes
> between linux-libc-dev_2.6.24-6 and linux-libc-dev_2.6.25-1
> are the first suspect.

OK, <asm/param.h> has changed on alpha like this:

--- 24/usr/include/asm/param.h 2008-04-22 11:49:18.000000000 +0300
+++ 25/usr/include/asm/param.h 2008-04-28 20:46:21.000000000 +0300
@@ -5,15 +5,7 @@
hardware ignores reprogramming. We also need userland buy-in to the
change in HZ, since this is visible in the wait4 resources etc. */

-
-#ifndef HZ
-# ifndef CONFIG_ALPHA_RAWHIDE
-# define HZ 1024
-# else
-# define HZ 1200
-# endif
-#endif
-
+#define HZ CONFIG_HZ
#define USER_HZ HZ

#define EXEC_PAGESIZE 8192

which causes the regression. Reassigning, please let me know if perl is
doing something wrong.

Cheers,
--
Niko Tyni ntyni@debian.org


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Thu, May 08, 2008 at 12:32:35PM +0300, Niko Tyni wrote:
>
> OK, <asm/param.h> has changed on alpha like this:
>
> --- 24/usr/include/asm/param.h 2008-04-22 11:49:18.000000000 +0300
> +++ 25/usr/include/asm/param.h 2008-04-28 20:46:21.000000000 +0300
> @@ -5,15 +5,7 @@
> hardware ignores reprogramming. We also need userland buy-in to the
> change in HZ, since this is visible in the wait4 resources etc. */
>
> -
> -#ifndef HZ
> -# ifndef CONFIG_ALPHA_RAWHIDE
> -# define HZ 1024
> -# else
> -# define HZ 1200
> -# endif
> -#endif
> -
> +#define HZ CONFIG_HZ
> #define USER_HZ HZ
>
> #define EXEC_PAGESIZE 8192

we won't diverge on linux-libc-dev headers from upstream,
please take that upstream to lkml.

i'd think Perl has no business into looking at the HZ value,
as it gives no sense at all now nowadays with the NO_HZ aka
dynticks usage.

--
maks



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Thu, May 08, 2008 at 12:01:20PM +0200, maximilian attems wrote:
> On Thu, May 08, 2008 at 12:32:35PM +0300, Niko Tyni wrote:
> >
> > OK, <asm/param.h> has changed on alpha like this:

> > +#define HZ CONFIG_HZ

FWIW, this was

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=bdc807871d58285737d50dc6163 d0feb72cb0dc2

A bit of googling (and a look in the code) indicates at least iputils
and python2.5 broke too.

http://bugs.gentoo.org/show_bug.cgi?id=218858

> we won't diverge on linux-libc-dev headers from upstream,
> please take that upstream to lkml.

I really think that's your domain, but if you insist.
Does something like the m68k fix proposed at

http://www.ussg.iu.edu/hypermail/linux/kernel/0804.3/3860.html

look right?

> i'd think Perl has no business into looking at the HZ value,
> as it gives no sense at all now nowadays with the NO_HZ aka
> dynticks usage.

Perl doesn't really use it as long as sysconf(_SC_CLK_TCK) works.
It also has a fallback if HZ isn't defined, the failure mode just
defeats that.

#ifndef HZ
# ifdef CLK_TCK
# define HZ CLK_TCK
# else
# define HZ 60
# endif
#endif

[...]

#if defined(HAS_SYSCONF) && defined(_SC_CLK_TCK) && !defined(__BEOS__)
PL_clocktick = sysconf(_SC_CLK_TCK);
if (PL_clocktick <= 0)
#endif
PL_clocktick = HZ;

--
Niko Tyni ntyni@debian.org



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Fri, May 09, 2008 at 10:22:53AM +0300, Niko Tyni wrote:

> > we won't diverge on linux-libc-dev headers from upstream,
> > please take that upstream to lkml.
>
> I really think that's your domain, but if you insist.

FWIW, there's a complete patch signed off by Mike Frysinger in

http://www.ussg.iu.edu/hypermail/linux/kernel/0805.1/1003.html

--
Niko Tyni ntyni@debian.org



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org