Re: [Xen-devel] no-amd-iommu-perdev-intremap + no-intremap = BOOM with Xen 4.4 (no-intremap by itself OK).

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 02.12.13 at 17:59, Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx> wrote:
> But if I also add "no-amd-iommu-perdev-intremap" (in addition to
> no-intremap) it blows up:

That option should really go away sooner or later anyway.

> (XEN) ----[ Xen-4.4-unstable  x86_64  debug=y  Not tainted ]----
> (XEN) CPU:    0
> (XEN) RIP:    e008:[<ffff82d0801285de>] check_lock+0x1e/0x3b
> (XEN) RFLAGS: 0000000000010002   CONTEXT: hypervisor
> (XEN) rax: 0000000000000001   rbx: ffff830448098c24   rcx: 0000000000000001
> (XEN) rdx: 0000000000000001   rsi: 0000000000000000   rdi: ffff830448098c28
> (XEN) rbp: ffff82d0802cfde8   rsp: ffff82d0802cfde8   r8:  ffff8301d143d200
> (XEN) r9:  0000000000000013   r10: ffff82d08023d160   r11: 0000000000000202
> (XEN) r12: 0000000000000286   r13: ffff8301d143d200   r14: ffff830448098c00
> (XEN) r15: 0000000000000000   cr0: 000000008005003b   cr4: 00000000000406f0
> (XEN) cr3: 0000000421c0c000   cr2: ffff830448098c28
> (XEN) ds: 0000   es: 0000   fs: 0000   gs: 0000   ss: e010   cs: e008
> (XEN) Xen stack trace from rsp=ffff82d0802cfde8:
> (XEN)    ffff82d0802cfe08 ffff82d0801286ec 000000000018298c ffff830448098c24
> (XEN)    ffff82d0802cfe48 ffff82d08016c1c5 0000000000000001 ffff8301d143d200
> (XEN)    ffff8300bcffc000 ffff880427403ecc 000000000000003e 0000000000000013
> (XEN)    ffff82d0802cfe68 ffff82d08016c59d ffff82d0802cfe68 ffff8301d143d200
> (XEN)    ffff82d0802cfef8 ffff82d080180249 ffff83042cbe4c00 ffff82d0802cfe90
> (XEN)    ffff82d0802e7f00 ffff82d0802e7f00 ffff82d0802c8000 ffffffff00000013
> (XEN)    0000000000000002 ffff82d0802cfee0 ffff82d080128019 ffff82d0802c8000
> (XEN)    ffff82d0802c8000 ffff8300bcffc000 000000000000003e 4000000000000000
> (XEN)    000000000000003e 0000000000000013 00007d2f7fd300c7 ffff82d0802248db
> (XEN)    ffffffff8100142a 0000000000000021 ffffffff81c01fd8 ffffffff81c01fd8
> (XEN)    ffffffff81c01fd8 0000000000000000 ffff880427403ee0 ffff880404e82f00
> (XEN)    0000000000000202 0000000000000015 ffff880410800000 ffff880404e82f00
> (XEN)    0000000000000021 ffffffff8100142a 0000000000000000 ffff880427403ecc
> (XEN)    000000000000000c 0001010000000000 ffffffff8100142a 000000000000e033
> (XEN)    0000000000000202 ffff880427403eb0 000000000000e02b 0000000000000000
> (XEN)    0000000000000000 0000000000000000 0000000000000000 0000000000000000
> (XEN)    ffff8300bcffc000 0000000000000000 0000000000000000
> (XEN) Xen call trace:
> (XEN)    [<ffff82d0801285de>] check_lock+0x1e/0x3b
> (XEN)    [<ffff82d0801286ec>] _spin_lock_irqsave+0x17/0x84
> (XEN)    [<ffff82d08016c1c5>] pirq_spin_lock_irq_desc+0x38/0x80
> (XEN)    [<ffff82d08016c59d>] pirq_guest_eoi+0x1f/0x36
> (XEN)    [<ffff82d080180249>] do_physdev_op+0x13c/0x112b
> (XEN)    [<ffff82d0802248db>] syscall_enter+0xeb/0x145
> (XEN) 
> (XEN) Pagetable walk from ffff830448098c28:
> (XEN)  L4[0x106] = 00000000ba681063 ffffffffffffffff
> (XEN)  L3[0x011] = 0000000000000000 ffffffffffffffff
> (XEN) 
> (XEN) ****************************************
> (XEN) Panic on CPU 0:
> (XEN) FATAL PAGE FAULT
> (XEN) [error_code=0000]
> (XEN) Faulting linear address: ffff830448098c28
> (XEN) ****************************************
> (XEN) 
> (XEN) Reboot in five seconds...
> (XEN) Resetting with ACPI MEMORY or I/O RESET_REG.

So you've got a page fault on the 1:1 mapping of PA
0x448098c28 here (yet physical memory ends at
0x43f000000, so one could suspect a negative interrupt
number converted to a 32-bit unsigned and the used as
array index), but ...

> Attached is the full serial log - of the "bad" and "failed" log.
>...
> (XEN) Xen BUG at spinlock.c:48
>...
> (XEN) Xen BUG at spinlock.c:48

... two BUG() instances (indicating improper spin lock use, with
and without interrupt disabled) here? Which one are we to
hunt down now?

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel