tags 604869 + confirmed upstream fixed-upstream thanks [Replying to an old bugreport... Quoting in full...]
On 25.11.2010 14:48, Jakub Wilk wrote: > * Michael Tokarev <m...@tls.msk.ru>, 2010-11-25, 10:19: >>> #0 0xf7786425 in __kernel_vsyscall () >>> #1 0xf7166751 in raise (sig=6) at >>> ../nptl/sysdeps/unix/sysv/linux/raise.c:64 >>> #2 0xf7169b82 in abort () at abort.c:92 >>> #3 0x080889c5 in oom_check (size=4294836496) at qemu-malloc.c:30 >>> #4 qemu_malloc (size=4294836496) at qemu-malloc.c:54 >>> #5 0x081f55aa in qemu_sglist_init (qsg=0x91890a0, alloc_hint=536862737) >>> at /tmp/buildd/qemu-kvm-0.13.0+dfsg/dma-helpers.c:15 >>> #6 0x081ef32b in dma_buf_prepare (bm=0x91899ec, is_write=<value >>> optimized out>) at /tmp/buildd/qemu-kvm-0.13.0+dfsg/hw/ide/core.c:439 >>> #7 0x081ef4c5 in ide_read_dma_cb (opaque=0x91899ec, ret=0) at >>> /tmp/buildd/qemu-kvm-0.13.0+dfsg/hw/ide/core.c:600 >>> #8 0x081f51e8 in dma_bdrv_cb (opaque=0x9246460, ret=0) at >>> /tmp/buildd/qemu-kvm-0.13.0+dfsg/dma-helpers.c:94 >>> #9 0x0808e480 in bdrv_aio_bh_cb (opaque=0x92312a0) at block.c:2311 >>> #10 0x080c68c8 in qemu_bh_poll () at async.c:150 >>> #11 0x0806311b in main_loop_wait (nonblocking=0) at >>> /tmp/buildd/qemu-kvm-0.13.0+dfsg/vl.c:1317 >>> #12 0x080799b4 in kvm_main_loop () at >>> /tmp/buildd/qemu-kvm-0.13.0+dfsg/qemu-kvm.c:1710 >>> #13 0x08065ce3 in main_loop (argc=7, argv=0xffc66154, envp=0xffc66174) >>> at /tmp/buildd/qemu-kvm-0.13.0+dfsg/vl.c:1341 >>> #14 main (argc=7, argv=0xffc66154, envp=0xffc66174) at >>> /tmp/buildd/qemu-kvm-0.13.0+dfsg/vl.c:3075 >> >> Excellent... not. >> >>> [0] You can try e.g. this image: >>> http://cdimage.debian.org/cdimage/squeeze_di_beta1/kfreebsd-i386/iso-cd/debian-squeeze-di-beta1-kfreebsd-i386-businesscard.iso >> >> I don't think it's specific to debian or freebsd. > > I guess so, but I was not able to trigger the bug with Linux kernel (I didn't > try hard, though). > >> Can you describe how to hit the bug, -- how your vfat directory looks like, >> and what did you do in the vm to trigger it? > > OK, to trigger the bug I run: > > kvm -cdrom debian-squeeze-di-beta1-kfreebsd-i386-businesscard.iso -boot d > -hda fat:empty/ > > choose "Default install" and just wait a few seconds. Last messages in the > guest kernel log are: > > ad0: 504MB <QEMU HARDDISK 0.13.0> at ata0-master WDMA2 > ad0: FAILURE - READ_DMA status=41<READY,ERROR> error=4<ABORTED> LBA=1032189 > > > "empty" is an empty directory on a partition like this: > > $ df -T empty/ > Filesystem Type 1K-blocks Used Available Use% Mounted on > /dev/mapper/root-crypt > ext4 144182892 95657860 48525032 67% / > > $ mount | grep root-crypt > /dev/mapper/root-crypt on / type ext4 > (rw,noatime,errors=remount-ro,acl,barrier=1,data=ordered) I'm not sure how I overlooked this bugreport when you provided a way to reproduce it. So acting on it now. I d/loaded current 6.0.4 kfreebsd installer and verified the bug is actually present in 0.13 as you described. And I found out that it does not happen anymore in 0.14.0 and above, including the current 1.0 version. Upstream git log has some mentions of SIGSEGV and memory corruption fixes in vvfat driver and in block layer. So it looks like one of these fixed the issue. I'm tagging this bug as "fixed-upstream" for now, can you please verify that current version does actually fix the issue for you too? If yes, I'll close this bug for 0.14.0 version. Thank you! /mjt -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
