Re: [gentoo-dev] Re: glibc-2.13 news item?
2011/2/9 Ryan Hill dirtye...@gentoo.org
On Tue, 08 Feb 2011 09:52:55 +0100
Paweł Hajdan, Jr. phajdan...@gentoo.org wrote:
It seems that with glibc-2.13 there are some serious compatibility
issues. There are good warnings on the planet
(http://psykil.livejournal.com/340806.html), but not every ~arch user
reads the planet, so how about creating news item with detailed
instructions how to ensure smooth glibc-2.13 update or recover a hosed
system?
We've blocked prelink in the ebuild so if a user is able to sync they're
either not going to be affected or already have been.
Also I've the strong suspect that these changes:
* New optimized string functions for x86-64: strnlen (SSE2),
strcasecmp (SSE2, SSSE3, SSE4.2), strncasecmp (SSE2, SSSE3, SSE4.2)
Implemented by Ulrich Drepper.
Interact badly with strong optimizations like these
CFLAGS=-O2 -march=core2 -pipe
CFLAGS=${CFLAGS} -msse4.1 --param l1-cache-size=32 --param
l1-cache-line-size=64 --param l2-cache-size=2048 -mtune=core2 # native
CFLAGS=${CFLAGS} -fgcse-after-reload -fpredictive-commoning
-ftree-vectorize -funswitch-loops # O3 - -finline-functions -fipa-cp-clone
CFLAGS=${CFLAGS} -fgraphite-identity -floop-block -floop-interchange
-floop-strip-mine # graphite co (-fira-loop-pressure no good amd64)
using gcc-4.5.2
Since the upgrade I do get portage emerging text files .sh, .conf and such
as file of the exact same size but filled of \0, luckily most upgrade fails.
cpu is:
model name : Intel(R) Core(TM)2 Quad CPUQ8400 @ 2.66GHz
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm
constant_tsc arch_perfmon pebs bts rep_good nopl aperfmperf pni dtes64
monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 xsave lahf_lm dts
tpr_shadow vnmi flexpriority
I've masked glibc-2.13, also I'm not filing a bug, because of non-usual
CFLAGS and because I'm not totally sure is glibc related.
Re: [gentoo-dev] Re: glibc-2.13 news item?
On Wed, 9 Feb 2011 13:32:36 +0100 Francesco R viv...@gmail.com wrote: Since the upgrade I do get portage emerging text files .sh, .conf and such as file of the exact same size but filled of \0, luckily most upgrade fails. I've seen similar issue in bug #353907 [1] but there they blame coreutils and/or btrfs. Maybe you should take a look. [1] http://bugs.gentoo.org/show_bug.cgi?id=353907 -- Best regards, Michał Górny signature.asc Description: PGP signature
Re: [gentoo-dev] Re: glibc-2.13 news item?
2011/2/9 Michał Górny mgo...@gentoo.org On Wed, 9 Feb 2011 13:32:36 +0100 Francesco R viv...@gmail.com wrote: Since the upgrade I do get portage emerging text files .sh, .conf and such as file of the exact same size but filled of \0, luckily most upgrade fails. I've seen similar issue in bug #353907 [1] but there they blame coreutils and/or btrfs. Maybe you should take a look. [1] http://bugs.gentoo.org/show_bug.cgi?id=353907 thanks Michał Górny, I had portage tmpdir on btrfs. My fault then, please ignore the noise in the my previous message. Best regards, Francesco R.
[gentoo-dev] Re: glibc-2.13 news item?
Francesco R posted on Wed, 09 Feb 2011 13:32:36 +0100 as excerpted: 2011/2/9 Ryan Hill dirtye...@gentoo.org On Tue, 08 Feb 2011 09:52:55 +0100 Paweł Hajdan, Jr. phajdan...@gentoo.org wrote: It seems that with glibc-2.13 there are some serious compatibility issues. [H]ow about creating news item with detailed instructions how to ensure smooth glibc-2.13 update or recover a hosed system? We've blocked prelink in the ebuild so if a user is able to sync they're either not going to be affected or already have been. Also I've the strong suspect that these changes: * New optimized string functions for x86-64: strnlen (SSE2), strcasecmp (SSE2, SSSE3, SSE4.2), strncasecmp (SSE2, SSSE3, SSE4.2) Implemented by Ulrich Drepper. Interact badly with strong optimizations like [snip] FYI... If you aren't deliberately modding the glibc ebuild, disabling its strippers, it strips most CFLAGS specifically for glibc only, due to problems they've had in the past with over-optimization and because if it breaks, the whole system does, so they want to be somewhat conservative. Basically, with the glibc ebuild you're stripped down to the -march/mtune setting, -O2 and -pipe. Pretty much everything else is stripped, as I said, unless you're running a customized ebuild with the stripping code disabled. As you didn't mention you were, I'll assume you aren't. You should of course be able to check your logs and see what CFLAGS were actually used in the compile, if there's any doubt. Of course, if you are modding the ebuild, disabling the stripping, then yes, for sure, don't file bugs on it! But you didn't /say/ you modified it, and people who disable that stripping aren't the sorts of folks the news items are written for anyway as they're /entirely/ on their own, so it's not like that'd be a concern, here. -- Duncan - List replies preferred. No HTML msgs. Every nonfree program has a lord, a master -- and if you use the program, he is your master. Richard Stallman
[gentoo-dev] Re: glibc-2.13 news item?
On Tue, 08 Feb 2011 09:52:55 +0100 Paweł Hajdan, Jr. phajdan...@gentoo.org wrote: It seems that with glibc-2.13 there are some serious compatibility issues. There are good warnings on the planet (http://psykil.livejournal.com/340806.html), but not every ~arch user reads the planet, so how about creating news item with detailed instructions how to ensure smooth glibc-2.13 update or recover a hosed system? We've blocked prelink in the ebuild so if a user is able to sync they're either not going to be affected or already have been. -- fonts, gcc-porting, it makes no sense how it makes no sense toolchain, wxwidgets but i'll take it free anytime @ gentoo.orgEFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662 signature.asc Description: PGP signature
