|
From: Julian S. <js...@ac...> - 2016-10-03 12:14:48
|
A beta tarball for 3.12.0 is now available at http://www.valgrind.org/downloads/valgrind-3.12.0.BETA1.tar.bz2 (md5sum = cb91572285f0072861a23ed8531b1829) Please give it a try in configurations that are important for you, and report any problems you have, either on this mailing list, or (preferably) via our bug tracker at https://bugs.kde.org/enter_bug.cgi?product=valgrind I will watch commits to the trunk and merge to 3.12.0, any that appear to be suitable candidates. If you know of any specific commits which you think should be merged, please mail the list to say so. If nothing critical emerges, a final release will happen on Thursday 20 October. Details of what's new in 3.12.0 will be in the NEWS file, although that is somewhat incomplete at present. Some of the highlights are: - Memcheck: replacement/wrapping of malloc/new related functions in all libraries, not just libc. - Memcheck: meta-mempool support for describing custom allocators - Valgrind is able to read compressed debuginfo sections - Modest JIT-cost improvements: 10%-15% for Memcheck on x86_64 - Support for IBM POWER ISA 3.0 instructions - Support for ARMv8 crypto instructions - Preliminary support for macOS 10.12 (Sierra) - Linux: support for recent glibc, gcc and kernel (a la Fedora 25) - Many bug fixes J |
|
From: Mark W. <mj...@re...> - 2016-10-03 12:25:40
|
On Mon, 2016-10-03 at 14:14 +0200, Julian Seward wrote: > I will watch commits to the trunk and merge to 3.12.0, any that > appear to be suitable candidates. If you know of any specific > commits which you think should be merged, please mail the list to > say so. For what it is worth I think you can fairly safely merge everything done on trunk since the branch was created. But if you want to be a little more careful then here is my review of the patches so far: - VEX There have been no commits to VEX since the branch was created. But there are some powerpc changes brewing in https://bugs.kde.org/show_bug.cgi?id=369175 The first seems good to go and I have already been testing it in the Fedora builds. The second with updates for Power9 BE should probably also go in once Carl finishes his test run. - valgrind - r15974 NEWS bugs update - Should be merged onto branch. - r15975 Avoid unused variable warning. - Harmless, so should be merged. - r15976 Preliminary support for macOS Sierra (10.12). - Unknown. - r15977 Powerpc big endian, fix the expected output file - Fixed .exp file should be merged. - r15978 Use proper compiler flags on Solaris for fma4 test. r15980 Fix none/tests/amd64/Makefile.am typo s/AM_CFKAGS/AM_CFLAGS/ - Bad compile flags (for the same test). Should be merged. - r15979 Fix expected error output of drd/tests/bar_bad* on Solaris. - Sigh. bar_bad... Should be merged. - r15981 Solaris: Add ioctl wrapper for MNTIOC_GETEXTMNTENT. - Unknown. - r15982 bug #361615 - Inconsistent termination for multithreaded process r15983 Fix warning introduced by revision 15982 - Looks good to me, very small risk, comes with testcase. - r15984 Added meta mempool support into memcheck for describing a custom allocator r15985 Fix test so that leaked bytes is the same in 32 and 64 bits r15986 Add an optional 2nd arg to leak-autofreepool to test performance r15987 mc-manual.xml: Fix some mismatched open/close tags. - New functionality that I haven't reviewed much, but seems low risk. - r15988 Use AM_LDFLAGS instead of LDFLAGS in exp-bbv/tests Makefiles. - Probably only really useful to me in Fedora packaging. - r15989 Don't require the current working directory to exist. Bug #369209. - Slight risk since it changes behavior. But comes with testcase and the new code should only trigger in situations where the old valgrind would go into an infinite loop. - r15990 Fix pre_mem_read_sockaddr crash on invalid syscall arguments. r15991 Fix crash in msghdr_foreachfield when iov_len isn't safe to deref r15992 Fix crash when old/new sigprocmask isn't safe to dereference. r15993 Fix crash in vmsplice linux kernel wrapper when iovec is bad. r15994 Fix crash in linux [rt_]sigaction wrapper with bad old/new sigaction r15995 Fix crash in sys_modify_ldt wrapper on bad ptr. r15997 Don't check bad iovec array in process_vm_readv/writev. r15998 Don't crash, but warn and return EINVAL on unknown fcntl command. - Various corner cases where valgrind would crash in the syscall wrappers found by LTP. Should be low risk. But probably don't happen much. It is useful in some cases when e.g. erroneously passing -1 instead of NULL to a syscall. Valgrind will notice by crashing, but getting a real report might be more helpful. I had to double check the sigaction and sigprocmask code because those are also used internally and so we have to make sure not to flag cases that are triggered by coregrind. I believe I got all that right. But that might be a reason to pass for now and postpone. - r15999 Replace --wait-for-gdb=yes memory loop by a call to VG_(poll) r16000 Well, 5 seconds is too short for me to type a attach pid command - Lets ask Philippe. |
|
From: Philippe W. <phi...@sk...> - 2016-10-04 06:01:46
|
On Mon, 2016-10-03 at 14:25 +0200, Mark Wielaard wrote: > - r15999 Replace --wait-for-gdb=yes memory loop by a call to VG_(poll) > r16000 Well, 5 seconds is too short for me to type a attach pid command > - Lets ask Philippe. This is a very small risk, concerning a valgrind developer only functionality. So is ok to merge (but not a big deal if not merged). Thanks Philippe |
|
From: osjup <dam...@gm...> - 2016-10-07 20:40:49
|
Julian Seward-2 wrote > Details of what's new in 3.12.0 will be in the NEWS file, > although that is somewhat incomplete at present. Some of the > highlights are: > ... > - Preliminary support for macOS 10.12 (Sierra) Preliminary support means I should already be able to use it, with the 'risk' of having some bugs right? Or it's not yet the point? Because when I'm trying just to configure it on my 10.12 I get this error: > configure: error: Valgrind works on Darwin 10.x, 11.x, 12.x, 13.x, 14.x > and 15.x (Mac OS X 10.6/7/8/9/10/11) Not sure whether it was meant to be like that in this beta, or not, so I thought I would just let you know. -- View this message in context: http://valgrind.10908.n7.nabble.com/Valgrind-3-12-0-BETA-is-available-for-testing-tp56801p56860.html Sent from the Valgrind - Dev mailing list archive at Nabble.com. |
|
From: Rhys K. <rhy...@gm...> - 2016-10-08 14:24:42
|
On 7 October 2016 at 15:17, osjup <dam...@gm...> wrote: > Julian Seward-2 wrote > > Details of what's new in 3.12.0 will be in the NEWS file, > > although that is somewhat incomplete at present. Some of the > > highlights are: > > ... > > - Preliminary support for macOS 10.12 (Sierra) > > Preliminary support means I should already be able to use it, with the > 'risk' of having some bugs right? Or it's not yet the point? > Because when I'm trying just to configure it on my 10.12 I get this error: > > > configure: error: Valgrind works on Darwin 10.x, 11.x, 12.x, 13.x, 14.x > > and 15.x (Mac OS X 10.6/7/8/9/10/11) > > Not sure whether it was meant to be like that in this beta, or not, so I > thought I would just let you know. > > Thanks for this report Damian. The macOS 10.12 (Sierra) support patch is indeed not in BETA1, although the email cover note mentions it as coming in 3.12 final causing potential confusion. Julian, it appears that r15976 has not yet made its way to the 3_12_BRANCHes. Regards, Rhys > > -- > View this message in context: http://valgrind.10908.n7. > nabble.com/Valgrind-3-12-0-BETA-is-available-for-testing- > tp56801p56860.html > Sent from the Valgrind - Dev mailing list archive at Nabble.com. > > ------------------------------------------------------------ > ------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |
|
From: Andreas A. <ar...@li...> - 2016-10-11 16:02:37
|
Resending after (re-)subscribing to the mailing list... On Mon, Oct 03 2016, Julian Seward wrote: > I will watch commits to the trunk and merge to 3.12.0, any that > appear to be suitable candidates. If you know of any specific > commits which you think should be merged, please mail the list to > say so. I'd like the fix for Bug 369439 to be merged. It consists of the following commits by Christian Bornträger: VEX r3259 Valgrind r16027 Valgrind r16028 Valgrind r16029 The fix affects s390x only and should apply cleanly for 3.12. It adds support for some instructions that may be emitted by newer compiler versions. -- Andreas |
|
From: Julian S. <js...@ac...> - 2016-10-18 05:50:09
|
Hi Andreas, > I'd like the fix for Bug 369439 to be merged. It consists of the > following commits by Christian Bornträger: > > VEX r3259 > Valgrind r16027 > Valgrind r16028 > Valgrind r16029 Merged. Thanks for the fixes. J |
|
From: Julian S. <js...@ac...> - 2016-10-18 06:23:06
|
Rhys, Sorry for the delay on merging. I just merged r15976 into the 3.12 branch, as r16071. It wasn't clear to me to what extent it will be possible to merge future Sierra support patches into the branch after release, at low risk, and to what extent 15976 actually gives usable Sierra support right now. But on looking at 15976 it looks pretty harmless, so I merged it. J On 08/10/16 16:24, Rhys Kidd wrote: > On 7 October 2016 at 15:17, osjup <dam...@gm...> wrote: > >> Julian Seward-2 wrote >>> Details of what's new in 3.12.0 will be in the NEWS file, >>> although that is somewhat incomplete at present. Some of the >>> highlights are: >>> ... >>> - Preliminary support for macOS 10.12 (Sierra) >> >> Preliminary support means I should already be able to use it, with the >> 'risk' of having some bugs right? Or it's not yet the point? >> Because when I'm trying just to configure it on my 10.12 I get this error: >> >>> configure: error: Valgrind works on Darwin 10.x, 11.x, 12.x, 13.x, 14.x >>> and 15.x (Mac OS X 10.6/7/8/9/10/11) >> >> Not sure whether it was meant to be like that in this beta, or not, so I >> thought I would just let you know. >> >> > Thanks for this report Damian. > > The macOS 10.12 (Sierra) support patch is indeed not in BETA1, although the > email cover note mentions it as coming in 3.12 final causing potential > confusion. > > Julian, it appears that r15976 has not yet made its way to the > 3_12_BRANCHes. > > Regards, > Rhys > > >> >> -- >> View this message in context: http://valgrind.10908.n7. >> nabble.com/Valgrind-3-12-0-BETA-is-available-for-testing- >> tp56801p56860.html >> Sent from the Valgrind - Dev mailing list archive at Nabble.com. >> >> ------------------------------------------------------------ >> ------------------ >> Check out the vibrant tech community on one of the world's most >> engaging tech sites, SlashDot.org! http://sdm.link/slashdot >> _______________________________________________ >> Valgrind-developers mailing list >> Val...@li... >> https://lists.sourceforge.net/lists/listinfo/valgrind-developers >> > > > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > > > > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |
|
From: Christian B. <bor...@de...> - 2016-10-18 10:26:36
|
Julian, after you have picked several commits (all look good, thanks), are you going to send out a new beta or rc tarball before the release? Christian |
|
From: Julian S. <js...@ac...> - 2016-10-20 14:24:47
|
On 18/10/16 09:06, Christian Borntraeger wrote: > Julian, > > after you have picked several commits (all look good, thanks), are you > going to send out a new beta or rc tarball before the release? Yes, I have made a RC2 tarball. I'll send out details in a separate message. J |