|
From: Randall H. <lis...@ch...> - 2003-11-10 14:48:32
|
Updated to valgrind from CVS this weekend, and the current emits errors for just about any program now. I backed up to the 20031012 snap and everything works. Here's the error from 20031108. This is on SuSE 8.2 BTW. ---------------------------------------------------------------------------- > /opt/pkg/valgrind-cvs-20031108/bin/valgrind ls ... ==5160== Memcheck, a.k.a. Valgrind, a memory error detector for x86-linux. ==5160== Copyright (C) 2002-2003, and GNU GPL'd, by Julian Seward. ==5160== Using valgrind-20030725, a program supervision framework for x86-linux. ==5160== Copyright (C) 2000-2003, and GNU GPL'd, by Julian Seward. ==5160== Estimated CPU clock rate is 1203 MHz ==5160== For more details, rerun with: -v ==5160== <ls output> TID 1: proxy LWP 0 exited with status 0 valgrind: vg_proxylwp.c:1331 (vgPlain_proxy_sanity): Assertion `sane' failed. ==5160== at 0x4016D5FB: vgPlain_skin_assert_fail (vg_mylibc.c:1103) ==5160== by 0x4016D5FA: assert_fail (vg_mylibc.c:1099) ==5160== by 0x4016D655: vgPlain_core_assert_fail (vg_mylibc.c:1110) ==5160== by 0x40170E2A: vgPlain_proxy_sanity (vg_proxylwp.c:1331) sched status: Thread 1: status = WaitSys, associated_mx = 0x0, associated_cv = 0x0 ==5160== at 0x4031A3A8: __GI___libc_write (in /lib/libc.so.6) ==5160== by 0x402B554F: new_do_write (in /lib/libc.so.6) ==5160== by 0x402B54ED: _IO_do_write@@GLIBC_2.1 (in /lib/libc.so.6) ==5160== by 0x402B5FF7: _IO_file_overflow@@GLIBC_2.1 (in /lib/libc.so.6) ---------------------------------------------------------------------------- Note: see also the FAQ.txt in the source distribution. It contains workarounds to several common problems. If that doesn't help, please report this bug to: js...@ac... In the bug report, send all the above text, the valgrind version, and what Linux distro you are using. Thanks. Randall |
|
From: Josef W. <Jos...@gm...> - 2003-11-10 15:08:23
|
On Monday 10 November 2003 15:48, Randall Hopper wrote: > Updated to valgrind from CVS this weekend, and the current emits errors for > just about any program now. I backed up to the 20031012 snap and > everything works. Here's the error from 20031108. > > This is on SuSE 8.2 BTW. Try --assume-2.4=yes. Josef |
|
From: Randall H. <lis...@ch...> - 2003-11-10 15:39:25
|
Josef Weidendorfer: |On Monday 10 November 2003 15:48, Randall Hopper wrote: |> Updated to valgrind from CVS this weekend, and the current emits errors for |> just about any program now. I backed up to the 20031012 snap and |> everything works. Here's the error from 20031108. |> |> This is on SuSE 8.2 BTW. | |Try --assume-2.4=yes. Where do I put it? valgrind (from CVS from this morning) won't take this on the command-line, configure doesn't take it, and I don't find the string "assume-2.4" anywhere in the source. Randall |
|
From: Nicholas N. <nj...@ca...> - 2003-11-10 15:49:27
|
On Mon, 10 Nov 2003, Randall Hopper wrote: > |Try --assume-2.4=yes. > > Where do I put it? Command-line, eg. valgrind --assume-2.4=yes myprogram > valgrind (from CVS from this morning) won't take this on the command-line, > configure doesn't take it, and I don't find the string "assume-2.4" > anywhere in the source. Did you check out the HEAD? From the new KDE repository (we moved away from SourceForge, although it hasn't been well publicised). But if you didn't use the KDE repository, I don't know how you're getting this problem... Either way, --assume-2.4=yes didn't fix the problem for me. N |
|
From: Randall H. <lis...@ch...> - 2003-11-11 22:43:04
|
Nicholas Nethercote: |Randall Hopper: |> |Try --assume-2.4=yes. | |> valgrind (from CVS from this morning) won't take this on the command-line, |> configure doesn't take it, and I don't find the string "assume-2.4" |> anywhere in the source. | |Did you check out the HEAD? From the new KDE repository (we moved away |from SourceForge, although it hasn't been well publicised). But if you |didn't use the KDE repository, I don't know how you're getting this |problem... Thanks. I'd pulled from sourceforge. Switching to KDE, --assume-2.4=yes gets me past that error, but then after some minutes valgrinding this app locks my machine hard; it's completely net dead and console dead. Don't see this with my 9/23 CVS snapshot. BTW, this is on SuSE 8.2. Randall |
|
From: Jeremy F. <je...@go...> - 2003-11-11 23:20:11
|
On Tue, 2003-11-11 at 14:43, Randall Hopper wrote: > Switching to KDE, --assume-2.4=yes gets me past that error, but then after > some minutes valgrinding this app locks my machine hard; it's completely > net dead and console dead. Don't see this with my 9/23 CVS snapshot. > > BTW, this is on SuSE 8.2. Ah, well, that's what we like to call a kernel bug. No matter what junk Valgrind does, it shouldn't lock the kernel like that. Obviously it could be a Valgrind bug too, and we'd like to fix it, but I wonder if there's a kernel update from SuSE, or perhaps you could try just installing a stock 2.4 kernel? What does your app do? Does it have lots of threads? Do lots of syscalls? Does it seem to be all working fine, and then it all falls over, or does it gradually collapse? Any oopses or other kernel messages on the console? Thanks, J |
|
From: Randall H. <lis...@ch...> - 2003-11-14 14:35:17
|
Jeremy Fitzhardinge: |On Tue, 2003-11-11 at 14:43, Randall Hopper wrote: |> Switching to KDE, --assume-2.4=yes gets me past that error, but then after |> some minutes valgrinding this app locks my machine hard; it's completely |> net dead and console dead. Don't see this with my 9/23 CVS snapshot. |> |> BTW, this is on SuSE 8.2. | |Ah, well, that's what we like to call a kernel bug. No matter what junk |Valgrind does, it shouldn't lock the kernel like that. | |Obviously it could be a Valgrind bug too, and we'd like to fix it, but I |wonder if there's a kernel update from SuSE, or perhaps you could try |just installing a stock 2.4 kernel? SuSE does have a kernel update, which I'm running. I can try a stock kernel later. |What does your app do? Does it have lots of threads? Do lots of |syscalls? Does it seem to be all working fine, and then it all falls |over, or does it gradually collapse? Any oopses or other kernel |messages on the console? The app has about 10 threads, one using OpenGL. Probably a fair. It seems to be working fine, albeit slowly compared to previous valgrind snaps and eventually the GUI just freezes and it quits responding to events. Same with keyboard, net, etc. xosview doesn't suggest that we're running out of memory -- I have a Gig on this box (though this valgrind snap was consuming more than previous snaps) . No other clues that I'm aware of. The ATI OpenGL drivers do use SSE which valgrind doesn't quite deal with right yet, but I was only debugging app startup which used to work fine with older snaps. Sorry I can't provide more useful input. I'm not really sure what to look for. Randall |
|
From: Jeremy F. <je...@go...> - 2003-11-14 16:52:23
|
On Fri, 2003-11-14 at 06:35, Randall Hopper wrote: > The app has about 10 threads, one using OpenGL. Probably a fair. It seems > to be working fine, albeit slowly compared to previous valgrind snaps and > eventually the GUI just freezes and it quits responding to events. Same > with keyboard, net, etc. xosview doesn't suggest that we're running out of > memory -- I have a Gig on this box (though this valgrind snap was consuming > more than previous snaps) . No other clues that I'm aware of. The ATI > OpenGL drivers do use SSE which valgrind doesn't quite deal with right yet, > but I was only debugging app startup which used to work fine with older > snaps. > > Sorry I can't provide more useful input. I'm not really sure what to look > for. Ah, you're using the binary-only ATI driver? I wouldn't be surprised if this were the culprit. Is there another driver which will work with your hardware? Could you try that? Also, while you're running your app, can you switch to the console (alt-ctrl-f1) and wait for the crash, so you can see if any kernel messages appear. That would give a good clue about what's crashing. J |