|
From: Julian S. <js...@ac...> - 2016-10-20 14:35:46
|
A release candidate tarball for 3.12.0 is now available for testing, at http://valgrind.org/downloads/valgrind-3.12.0.RC2.tar.bz2 (md5sum = b057a598b2e7298918308dfc6903c141) Please give it a try in configurations that are important to you, and report any problems (and also successes!) to this list. If no critical problems emerge I'll create the 3.12.0 final tarball tomorrow afternoon. Thanks. J |
|
From: Christian B. <bor...@de...> - 2016-10-20 18:47:55
|
On 10/20/2016 04:35 PM, Julian Seward wrote: > > A release candidate tarball for 3.12.0 is now available for testing, at > > http://valgrind.org/downloads/valgrind-3.12.0.RC2.tar.bz2 > (md5sum = b057a598b2e7298918308dfc6903c141) > > Please give it a try in configurations that are important to you, and report > any problems (and also successes!) to this list. If no critical problems > emerge I'll create the 3.12.0 final tarball tomorrow afternoon. Looks ok for s390x on SLES12SP1. |
|
From: Mark W. <mj...@re...> - 2016-10-21 01:10:10
|
On Thu, 2016-10-20 at 16:35 +0200, Julian Seward wrote: > A release candidate tarball for 3.12.0 is now available for testing, at > > http://valgrind.org/downloads/valgrind-3.12.0.RC2.tar.bz2 > (md5sum = b057a598b2e7298918308dfc6903c141) > > Please give it a try in configurations that are important to you, and report > any problems (and also successes!) to this list. If no critical problems > emerge I'll create the 3.12.0 final tarball tomorrow afternoon. It is looking pretty good on RHEL7 (ppc, x86_64, ppc64le, ppc64, aarch64, i686 and s390x). On Fedora rawhide I hit bug 371396. I checked in a workaround as valgrind svn r16097. But you might find that too risky to add to the release branch (I only tested it on glibc based systems). At the moment the new pthread cond var implementation is Fedora rawhide only (the upcoming Fedora 25 is fine), so I don't mind having to carry a patch for it. Eventually it will go into glibc upstream at which point others might hit it. But that might take some time, so it can maybe wait till 3.12.1 in a couple of months. Cheers, Mark |
|
From: Petar J. <mip...@gm...> - 2016-10-21 22:08:57
|
Results from the tarball look fine on selected MIPS targets. Regards, Petar |
|
From: Marcin J. <mju...@re...> - 2016-10-22 09:08:14
|
W dniu 22.10.2016 o 00:08, Petar Jovanovic pisze: > Results from the tarball look fine on selected MIPS targets. If you have access to mips64 machine then can you copy PRE(fadvise64) from other 64bit arch (amd64, arm64, ppc64, s390x) and test? mips64 is the only 64-bit arch in valgrind which is using generic wrapper for this syscall. And generic one looks like written for 32bit targets. |
|
From: Petar J. <mip...@gm...> - 2016-11-29 15:00:17
|
On Sat, Oct 22, 2016 at 11:08 AM, Marcin Juszkiewicz <mju...@re...> wrote: > If you have access to mips64 machine then can you copy PRE(fadvise64) > from other 64bit arch (amd64, arm64, ppc64, s390x) and test? > > mips64 is the only 64-bit arch in valgrind which is using generic > wrapper for this syscall. And generic one looks like written for 32bit > targets. > Thank you for noticing this. Change committed in r16163. Regards, Petar |