Re: [Qemu-devel] [PATCH] block: add the support to drain throttled requests

[Stefan Hajnoczi]

On Fri, Mar 23, 2012 at 10:43 AM, Chris Webb <address@hidden> wrote:
> Stefan Hajnoczi <address@hidden> writes:
>
>> On Thu, Mar 22, 2012 at 07:07:52PM +0000, Chris Webb wrote:
>> > Stefan Hajnoczi <address@hidden> writes:
>> >
>> > > Yesterday I only posted an analysis of the bug but here are some
>> > > thoughts on how to move forward.  Throttling itself is not the problem.
>> > > We've known that synchronous operations in the vcpu thread are a problem
>> > > long before throttling.  This is just another reason to convert device
>> > > emulation to use asynchronous interfaces.
>> > >
>> > > Here is the list of device models that perform synchronous block I/O:
>> > > hw/fdc.c
>> > > hw/ide/atapi.c
>> > > hw/ide/core.c
>> > > hw/nand.c
>> > > hw/onenand.c
>> > > hw/pflash_cfi01.c
>> > > hw/pflash_cfi02.c
>> > > hw/sd.c
>> > >
>> > > Zhi Hui Li is working on hw/fdc.c and recently sent a patch.
>> > >
>> > > I think it's too close to QEMU 1.1 to convert all the remaining devices
>> > > and test them properly before the soft-freeze.  But it's probably
>> > > possible to convert IDE before the soft-freeze.
>> >
>> > IDE is the only of these that would affect us as a typical user of
>> > throttling. The others aren't really the kind of devices which you'd be
>> > using in a hosting setting in any case.
>>
>> Can you check whether your Windows guest has DMA or PIO mode enabled?
>>
>> http://msdn.microsoft.com/en-us/windows/hardware/gg463526
>
> Hi. We were producing the IDE assert()s and deadlocks with linux kernels.
> Although I believe the same symptoms exist on windows, I haven't actually
> tested it myself. Typically they would show up in the 16-bit bootloader
> code, even before the 32-bit OS has started.

Okay, that makes sense.  Bootloaders and the BIOS may use the simplest
driver interface - which may be PIO in the case.  I asked because the
IDE DMA code path should work with I/O throttling and Windows is known
for sometimes falling back to the PIO code path when some heuristics
trigger.

Stefan