I tried this on PAE Xen on a machine with 8GB. # xm list Name ID Mem(MiB) VCPUs State Time(s) Domain-0 0 7964 1 r----- 137.9 Then, simply # xm mem-max 0 5000 # xm list Causes: kernel BUG at arch/xen/i386/mm/hypervisor.c:381 (xen_create_con tiguous_region)! [<c0119faa>] xen_create_contiguous_region+0x34a/0x4e0 [<c0116ec2>] pgd_ctor+0x22/0x130 [<c014b0a7>] cache_init_objs+0x67/0x70 [<c014b282>] cache_grow+0x122/0x210 [<c014b4e3>] cache_alloc_refill+0x173/0x250 [<c014b805>] kmem_cache_alloc+0x95/0xa0 [<c01170e3>] pgd_alloc+0x23/0x320 [<c0111301>] sync_single+0x51/0xa0 [<c011db23>] mm_init+0xa3/0xe0 [<c011de6e>] copy_mm+0xbe/0x3d0 [<c010484c>] force_evtchn_callback+0xc/0x10 [<c011ea97>] copy_process+0x3f7/0xd40 [<c011f4d9>] do_fork+0x69/0x184 [<c0223142>] copy_to_user+0x42/0x90 [<c0107b9c>] sys_clone+0x3c/0x40 [<c0109355>] syscall_call+0x7/0xb Kernel panic - not syncing: BUG! (XEN) Domain 0 shutdown: rebooting machine. I used: changeset: 7745:b9e8654c3f10 tag: tip user: kaf24@firebug.cl.cam.ac.uk date: Thu Nov 10 19:43:56 2005 +0100 summary: Fix Linux fault.c indentation. I tried samething on x86_64 xenlinux, but it worked fine. Jun --- Intel Open Source Technology Center _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
BTW, # xm mem-set 5000 worked fine. Looks like it''s already in http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=212 So we need to remove it from xm? BTW Xen on FC4 page (http://www.fedoraproject.org/wiki/FedoraXenQuickstart) suggests the user do the following (as an exmaple) xm mem-max 0 384 xm mem-set 0 384 Nakajima, Jun wrote:> I tried this on PAE Xen on a machine with 8GB. > > # xm list > Name ID Mem(MiB) VCPUs State Time(s) > Domain-0 0 7964 1 r----- 137.9 > > Then, simply > # xm mem-max 0 5000 > # xm list > > Causes: > > kernel BUG at arch/xen/i386/mm/hypervisor.c:381 (xen_create_con > tiguous_region)! > [<c0119faa>] xen_create_contiguous_region+0x34a/0x4e0 > [<c0116ec2>] pgd_ctor+0x22/0x130 > [<c014b0a7>] cache_init_objs+0x67/0x70 > [<c014b282>] cache_grow+0x122/0x210 > [<c014b4e3>] cache_alloc_refill+0x173/0x250 > [<c014b805>] kmem_cache_alloc+0x95/0xa0 > [<c01170e3>] pgd_alloc+0x23/0x320 > [<c0111301>] sync_single+0x51/0xa0 > [<c011db23>] mm_init+0xa3/0xe0 > [<c011de6e>] copy_mm+0xbe/0x3d0 > [<c010484c>] force_evtchn_callback+0xc/0x10 > [<c011ea97>] copy_process+0x3f7/0xd40 > [<c011f4d9>] do_fork+0x69/0x184 > [<c0223142>] copy_to_user+0x42/0x90 > [<c0107b9c>] sys_clone+0x3c/0x40 > [<c0109355>] syscall_call+0x7/0xb > Kernel panic - not syncing: BUG! > (XEN) Domain 0 shutdown: rebooting machine. > > I used: > changeset: 7745:b9e8654c3f10 > tag: tip > user: kaf24@firebug.cl.cam.ac.uk > date: Thu Nov 10 19:43:56 2005 +0100 > summary: Fix Linux fault.c indentation. > > I tried samething on x86_64 xenlinux, but it worked fine. > > Jun > --- > Intel Open Source Technology Center > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-develJun --- Intel Open Source Technology Center _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
> I tried this on PAE Xen on a machine with 8GB. > > # xm list > Name ID Mem(MiB) VCPUs State Time(s) > Domain-0 0 7964 1 r----- 137.9 > > Then, simply > # xm mem-max 0 5000 > # xm list > > Causes: > > kernel BUG at arch/xen/i386/mm/hypervisor.c:381 > (xen_create_con tiguous_region)!''This may possibly be fixed by changeset 7771 (in staging tree) Patch attached. Thanks, Ian> [<c0119faa>] xen_create_contiguous_region+0x34a/0x4e0 > [<c0116ec2>] pgd_ctor+0x22/0x130 > [<c014b0a7>] cache_init_objs+0x67/0x70 > [<c014b282>] cache_grow+0x122/0x210 > [<c014b4e3>] cache_alloc_refill+0x173/0x250 [<c014b805>] > kmem_cache_alloc+0x95/0xa0 [<c01170e3>] pgd_alloc+0x23/0x320 > [<c0111301>] sync_single+0x51/0xa0 [<c011db23>] > mm_init+0xa3/0xe0 [<c011de6e>] copy_mm+0xbe/0x3d0 > [<c010484c>] force_evtchn_callback+0xc/0x10 [<c011ea97>] > copy_process+0x3f7/0xd40 [<c011f4d9>] do_fork+0x69/0x184 > [<c0223142>] copy_to_user+0x42/0x90 [<c0107b9c>] > sys_clone+0x3c/0x40 [<c0109355>] syscall_call+0x7/0xb Kernel > panic - not syncing: BUG! > (XEN) Domain 0 shutdown: rebooting machine. > > I used: > changeset: 7745:b9e8654c3f10 > tag: tip > user: kaf24@firebug.cl.cam.ac.uk > date: Thu Nov 10 19:43:56 2005 +0100 > summary: Fix Linux fault.c indentation. > > I tried samething on x86_64 xenlinux, but it worked fine. > > Jun > --- > Intel Open Source Technology Center > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Ian Pratt wrote:>> I tried this on PAE Xen on a machine with 8GB. >> >> # xm list >> Name ID Mem(MiB) VCPUs State Time(s) >> Domain-0 0 7964 1 r----- 137.9 >> >> Then, simply >> # xm mem-max 0 5000 >> # xm list >> >> Causes: >> >> kernel BUG at arch/xen/i386/mm/hypervisor.c:381 >> (xen_create_con tiguous_region)!'' > > This may possibly be fixed by changeset 7771 (in staging tree) > Patch attached.Actually this causes a hang like: hda: TEAC DW-552G, ATAPI CD/DVD-ROM drive ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 hda: ATAPI 52X DVD-ROM CD-R/RW drive, 2048kB Cache, UDMA(33) Uniform CD-ROM driver Revision: 3.20 Red Hat/Adaptec aacraid driver (1.1.2-lk2 Nov 11 2005) 3ware Storage Controller device driver for Linux v1.26.02.001. ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 7 PCI: setting IRQ 7 as level-triggered ACPI: PCI Interrupt 0000:00:1f.2[B] -> Link [LNKD] -> GSI 7 (level, low) -> IRQ 7 ata1: SATA max UDMA/133 cmd 0x70C8 ctl 0x70E6 bmdma 0x70A0 irq 7 ata2: SATA max UDMA/133 cmd 0x70C0 ctl 0x70E2 bmdma 0x70A8 irq 7 ata1: dev 0 ATA, max UDMA/133, 234441648 sectors: lba48 ata1: dev 1 ATA, max UDMA/133, 234441648 sectors: lba48 All processors are in idle_loop in xen. Jun --- Intel Open Source Technology Center _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
> >> kernel BUG at arch/xen/i386/mm/hypervisor.c:381 (xen_create_con > >> tiguous_region)!'' > > > > This may possibly be fixed by changeset 7771 (in staging > tree) Patch > > attached. > > Actually this causes a hang like:I''d be pretty surprised if that patch caused the hang as it''s a pretty clear cut bug fix. Are you sure? Thanks, Ian> hda: TEAC DW-552G, ATAPI CD/DVD-ROM drive ide0 at > 0x1f0-0x1f7,0x3f6 on irq 14 > hda: ATAPI 52X DVD-ROM CD-R/RW drive, 2048kB Cache, UDMA(33) > Uniform CD-ROM driver Revision: 3.20 Red Hat/Adaptec aacraid > driver (1.1.2-lk2 Nov 11 2005) 3ware Storage Controller > device driver for Linux v1.26.02.001. > ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 7 > PCI: setting IRQ 7 as level-triggered > ACPI: PCI Interrupt 0000:00:1f.2[B] -> Link [LNKD] -> GSI 7 > (level, low) > -> IRQ > 7 > ata1: SATA max UDMA/133 cmd 0x70C8 ctl 0x70E6 bmdma 0x70A0 irq 7 > ata2: SATA max UDMA/133 cmd 0x70C0 ctl 0x70E2 bmdma 0x70A8 irq 7 > ata1: dev 0 ATA, max UDMA/133, 234441648 sectors: lba48 > ata1: dev 1 ATA, max UDMA/133, 234441648 sectors: lba48 > > > All processors are in idle_loop in xen. > > > Jun > --- > Intel Open Source Technology Center >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Ian Pratt wrote:>>>> kernel BUG at arch/xen/i386/mm/hypervisor.c:381 (xen_create_con >>>> tiguous_region)!'' >>> >>> This may possibly be fixed by changeset 7771 (in staging tree) Patch >>> attached. >> >> Actually this causes a hang like: > > I''d be pretty surprised if that patch caused the hang as it''s a pretty > clear cut bug fix. > Are you sure?Okay, I just remember this nasty issue with menuconfig. If you make SMP dom0, and go back to UP using memuconfig, the config file gets broken (CONFIG_APIC stuff). The UP dom0 does not boot anymore. Sorry, it worked fine.> > Thanks, > Ian > >> hda: TEAC DW-552G, ATAPI CD/DVD-ROM drive ide0 at >> 0x1f0-0x1f7,0x3f6 on irq 14 >> hda: ATAPI 52X DVD-ROM CD-R/RW drive, 2048kB Cache, UDMA(33) >> Uniform CD-ROM driver Revision: 3.20 Red Hat/Adaptec aacraid >> driver (1.1.2-lk2 Nov 11 2005) 3ware Storage Controller >> device driver for Linux v1.26.02.001. >> ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 7 >> PCI: setting IRQ 7 as level-triggered >> ACPI: PCI Interrupt 0000:00:1f.2[B] -> Link [LNKD] -> GSI 7 (level, >> low) -> IRQ >> 7 >> ata1: SATA max UDMA/133 cmd 0x70C8 ctl 0x70E6 bmdma 0x70A0 irq 7 >> ata2: SATA max UDMA/133 cmd 0x70C0 ctl 0x70E2 bmdma 0x70A8 irq 7 >> ata1: dev 0 ATA, max UDMA/133, 234441648 sectors: lba48 >> ata1: dev 1 ATA, max UDMA/133, 234441648 sectors: lba48 >> >> >> All processors are in idle_loop in xen. >> >> >> Jun >> --- >> Intel Open Source Technology Center >> > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-develJun --- Intel Open Source Technology Center _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
> > I''d be pretty surprised if that patch caused the hang as > it''s a pretty > > clear cut bug fix. > > Are you sure? > > Okay, I just remember this nasty issue with menuconfig. If > you make SMP dom0, and go back to UP using memuconfig, the > config file gets broken (CONFIG_APIC stuff). The UP dom0 does > not boot anymore.Yep, we should probably figure out how to make modify the Kconfig to make sure that one of CONFIG_X86_UP_IOAPIC and CONFIG_SMP is set. Ian _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 12 Nov 2005, at 03:01, Nakajima, Jun wrote:>>>>> kernel BUG at arch/xen/i386/mm/hypervisor.c:381 (xen_create_con >>>>> tiguous_region)!'' >>>> >>>> This may possibly be fixed by changeset 7771 (in staging tree) Patch >>>> attached. >>> >>> Actually this causes a hang like: >> >> I''d be pretty surprised if that patch caused the hang as it''s a pretty >> clear cut bug fix. >> Are you sure? > > Okay, I just remember this nasty issue with menuconfig. If you make SMP > dom0, and go back to UP using memuconfig, the config file gets broken > (CONFIG_APIC stuff). The UP dom0 does not boot anymore. > > Sorry, it worked fine.I''m surprised as the original bug you reported was actually a misuse of ''xm mem-max'' (which almost certainly should not be exported to users in its raw form). You set the Xen memory limit for that domain to less than its current allocation. Then, next time it tried to alloc a pgd, it tried to exchange the underlying mfn for one below 4GB. Xen took the old mfn away from the domain, but then failed the reallocation because the domain already had too much memory! ''xm mem-set'' is all that should be exported to users, and the hard limit should be under the control of xend. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel