Re: [Xen-devel] Double gnttab_end_access in mini-os netfront
On Sun, 2015-12-06 at 03:19 +0100, Marek Marczykowski-Górecki wrote: > Hi, > > When running HVM on Xen 4.6 with qemu in stubdom, I've found that > something goes wrong with disk frontend (at least that was visible > problem - a lot of read and write errors in stubdom log). But further > debugging (including -DGNT_DEBUG) leads to double gnttab_end_access in > netfront. > > The stack trace is: > ASSERTION FAILED: !(!inuse[ref]) at gnttab.c:42. > Do_exit called! > 0x000f3ffb: get_time_values_from_xen at xen-4.6.0/extras/mini- > os/arch/x86/time.c:134 (discriminator 1) > 0x000d74a8: HYPERVISOR_sched_op at xen-4.6.0/extras/mini- > os/include/x86/x86_64/hypercall-x86_64.h:180 > 0x000d6a2e: put_free_entry at xen-4.6.0/extras/mini- > os/gnttab.c:43 > 0x000d6bff: gnttab_end_access at xen-4.6.0/extras/mini- > os/gnttab.c:115 > 0x000d8a50: network_rx at xen-4.6.0/extras/mini-os/netfront.c:134 > 0x000d9ec4: netfront_receive at xen-4.6.0/extras/mini- > os/netfront.c:671 > 0x000dd302: get_current at xen-4.6.0/extras/mini- > os/include/x86/arch_sched.h:16 > 0x00079f72: tap_send at xen-4.6.0/stubdom/../tools/qemu-xen- > traditional/net.c:756 > 0x69f9: main_loop_wait at xen-4.6.0/stubdom/../tools/qemu- > xen-traditional/vl.c:3826 > 0x00021f27: main_loop at xen-4.6.0/stubdom/../tools/qemu-xen- > traditional/i386-dm/helper2.c:612 (discriminator 1) > 0x950d: quit_timers at xen-4.6.0/stubdom/../tools/qemu-xen- > traditional/vl.c:1866 > 0x000d7f57: call_main at xen-4.6.0/extras/mini-os/main.c:163 > 0x3423: thread_starter at :? > 0x: _start at ??:? > > It was working in Xen 4.4. The only commit touching xenfront.c (in any > meaningful way) from that time is this one: > http://xenbits.xen.org/gitweb/?p=mini-os.git;a=commit;h=7c8f348390652a67e > 9356eec9cd2b0f76a9c7c72 > > With that commit reverted, issue vanishes. > > I guess it's because before this commit, there was "if (rx->status == > NETIF_RSP_NULL) continue" before "gnttab_end_access(buf->gref)", but now > that case is handled after gnttab_end_access (using "if (rx->status > > NETIF_RSP_NULL)"). I think the fix would be to restore that "continue" > line. That sounds pretty plausible to me (FWIW). Have you tried it? > PS What is the correct place for such reports? minios-devel? xen-devel? > both? Formally I suppose the former, but realistically including xen-devel as well is likely to attract more eyes. Ian. ___ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
Re: [Xen-devel] Double gnttab_end_access in mini-os netfront
On Tue, Dec 08, 2015 at 11:33:49AM +, Ian Campbell wrote: > On Sun, 2015-12-06 at 03:19 +0100, Marek Marczykowski-Górecki wrote: > > Hi, > > > > When running HVM on Xen 4.6 with qemu in stubdom, I've found that > > something goes wrong with disk frontend (at least that was visible > > problem - a lot of read and write errors in stubdom log). But further > > debugging (including -DGNT_DEBUG) leads to double gnttab_end_access in > > netfront. > > > > The stack trace is: > > ASSERTION FAILED: !(!inuse[ref]) at gnttab.c:42. > > Do_exit called! > > 0x000f3ffb: get_time_values_from_xen at xen-4.6.0/extras/mini- > > os/arch/x86/time.c:134 (discriminator 1) > > 0x000d74a8: HYPERVISOR_sched_op at xen-4.6.0/extras/mini- > > os/include/x86/x86_64/hypercall-x86_64.h:180 > > 0x000d6a2e: put_free_entry at xen-4.6.0/extras/mini- > > os/gnttab.c:43 > > 0x000d6bff: gnttab_end_access at xen-4.6.0/extras/mini- > > os/gnttab.c:115 > > 0x000d8a50: network_rx at xen-4.6.0/extras/mini-os/netfront.c:134 > > 0x000d9ec4: netfront_receive at xen-4.6.0/extras/mini- > > os/netfront.c:671 > > 0x000dd302: get_current at xen-4.6.0/extras/mini- > > os/include/x86/arch_sched.h:16 > > 0x00079f72: tap_send at xen-4.6.0/stubdom/../tools/qemu-xen- > > traditional/net.c:756 > > 0x69f9: main_loop_wait at xen-4.6.0/stubdom/../tools/qemu- > > xen-traditional/vl.c:3826 > > 0x00021f27: main_loop at xen-4.6.0/stubdom/../tools/qemu-xen- > > traditional/i386-dm/helper2.c:612 (discriminator 1) > > 0x950d: quit_timers at xen-4.6.0/stubdom/../tools/qemu-xen- > > traditional/vl.c:1866 > > 0x000d7f57: call_main at xen-4.6.0/extras/mini-os/main.c:163 > > 0x3423: thread_starter at :? > > 0x: _start at ??:? > > > > It was working in Xen 4.4. The only commit touching xenfront.c (in any > > meaningful way) from that time is this one: > > http://xenbits.xen.org/gitweb/?p=mini-os.git;a=commit;h=7c8f348390652a67e > > 9356eec9cd2b0f76a9c7c72 > > > > With that commit reverted, issue vanishes. > > > > I guess it's because before this commit, there was "if (rx->status == > > NETIF_RSP_NULL) continue" before "gnttab_end_access(buf->gref)", but now > > that case is handled after gnttab_end_access (using "if (rx->status > > > NETIF_RSP_NULL)"). I think the fix would be to restore that "continue" > > line. > > That sounds pretty plausible to me (FWIW). Have you tried it? I've tried moving gnttab_end_access into that if branch. And it didn't worked. > > PS What is the correct place for such reports? minios-devel? xen-devel? > > both? > > Formally I suppose the former, but realistically including xen-devel as > well is likely to attract more eyes. > > Ian. -- Best Regards, Marek Marczykowski-Górecki Invisible Things Lab A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? pgpoNWwNfJQpJ.pgp Description: PGP signature ___ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
Re: [Xen-devel] Double gnttab_end_access in mini-os netfront
Ian Campbell, on Tue 08 Dec 2015 11:33:49 +, wrote: > > http://xenbits.xen.org/gitweb/?p=mini-os.git;a=commit;h=7c8f348390652a67e > > 9356eec9cd2b0f76a9c7c72 > > > > With that commit reverted, issue vanishes. > > > > I guess it's because before this commit, there was "if (rx->status == > > NETIF_RSP_NULL) continue" before "gnttab_end_access(buf->gref)", but now > > that case is handled after gnttab_end_access (using "if (rx->status > > > NETIF_RSP_NULL)"). I think the fix would be to restore that "continue" > > line. > > That sounds pretty plausible to me (FWIW). Have you tried it? That looks right, indeed. Samuel ___ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
Re: [Xen-devel] Double gnttab_end_access in mini-os netfront
Marek Marczykowski-Górecki, on Tue 08 Dec 2015 12:46:31 +0100, wrote: > > > http://xenbits.xen.org/gitweb/?p=mini-os.git;a=commit;h=7c8f348390652a67e > > > 9356eec9cd2b0f76a9c7c72 > > > > > > With that commit reverted, issue vanishes. > > > > > > I guess it's because before this commit, there was "if (rx->status == > > > NETIF_RSP_NULL) continue" before "gnttab_end_access(buf->gref)", but now > > > that case is handled after gnttab_end_access (using "if (rx->status > > > > NETIF_RSP_NULL)"). I think the fix would be to restore that "continue" > > > line. > > > > That sounds pretty plausible to me (FWIW). Have you tried it? > > I've tried moving gnttab_end_access into that if branch. And it didn't > worked. Which if branch? Please show the code, C is less ambiguous than english :) Samuel ___ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
