Re: [Qemu-devel] [PATCH for 2.6 v4 0/3] ide: fix loss of the dma/atapi state during migration

[John Snow]

On 04/12/2016 01:33 AM, Denis V. Lunev wrote:
> On 04/12/2016 12:47 AM, John Snow wrote:
>>
>> On 04/11/2016 10:12 AM, Denis V. Lunev wrote:
>>> On 04/06/2016 09:40 AM, Denis V. Lunev wrote:
>>>> This patch set fixes bugs in the IDE DMA and the IDE ATAPI on
>>>> operations to
>>>> save/restore the state.
>>>>
>>>>   From the user point of view this results in IDE timeouts in the guest
>>>> when the user reads from the DVD like the following:
>>>>
>>>> [424332.169229] ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action
>>>> 0x6 frozen
>>>> [424332.170423] sr 0:0:0:0: [sr0] CDB:
>>>> [424332.171234] Read(10): 28 00 00 00 02 e4 00 00 01 00
>>>> [424332.172418] ata1.00: cmd a0/01:00:00:00:08/00:00:00:00:00/a0 tag 0
>>>> dma 2048 in
>>>>            res 40/00:02:00:0c:00/00:00:00:00:00/a0 Emask 0x4 (timeout)
>>>> [424332.174877] ata1.00: status: { DRDY }
>>>> [424337.212099] ata1: link is slow to respond, please be patient
>>>> (ready=0)
>>>> [424342.220084] ata1: device not ready (errno=-16), forcing hardreset
>>>> [424342.222700] ata1: soft resetting link
>>>> [424342.381059] ata1.00: configured for MWDMA2
>>>> [424342.383693] ata1: EH complete
>>>>
>>>> Another similar nasty effects are possible.
>>>>
>>>> Signed-off-by: Pavel Butsykin <address@hidden>
>>>> Reviewed-by: Roman Kagan <address@hidden>
>>>> Signed-off-by: Denis V. Lunev <address@hidden>
>>>> CC: John Snow <address@hidden>
>>>>
>>>> Changes from v1:
>>>> - added converter of IDE_DMA_* to IDE_RETRY_* (1)
>>>> - fixed handling of the IDE_RETRY_HBA at the ide_restart_bh function
>>>> (3)
>>>>
>>>> Changes from v2:
>>>> - fixed enumeration value ‘IDE_DMA_ATAPI’ not handled in switch for
>>>> macio.c (3)
>>>>
>>>> Changes from v3:
>>>> - move the IDE_DMA_ATAPI setting in the cmd_packet func (3)
>>>>
>>>> Pavel Butsykin (3):
>>>>     ide: don't lose pending dma state
>>>>     ide: restart atapi dma by re-evaluating command packet
>>>>     ide: really restart pending and in-flight atapi dma
>>>>
>>>>    hw/ide/atapi.c    | 26 ++++++++++++--------------
>>>>    hw/ide/core.c     | 39 ++++++++++++++++-----------------------
>>>>    hw/ide/internal.h | 36 ++++++++++++++++++++++++++++++++++++
>>>>    hw/ide/macio.c    |  2 ++
>>>>    hw/ide/pci.c      |  4 ++++
>>>>    5 files changed, 70 insertions(+), 37 deletions(-)
>>>>
>>> ping
>> Sorry, I don't appear to have been CC'd on the actual patch emails, so I
>> missed them. The individual patch emails also miss the "v4" tag which
>> makes them hard to spot.... and it's hard to figure out what's been
>> changed or not since my reviewed-by was added to all three patches, even
>> though I only acknowledged the first 2/3.an reaction time of about 265 
>> millise
>>
> this is strange for me - you R-b: was added only to patches 1-2, here is
> a quote
> 
> "The patch fixes the restart of both in-flight and pending ATAPI DMA,
> following the scheme similar to that of IDE DMA.
> 
> Signed-off-by: Pavel Butsykin<address@hidden>
> Signed-off-by: Denis V. Lunev<address@hidden>
> ---
>  hw/ide/atapi.c    | 13 ++++++-------
>  hw/ide/core.c     | 30 +++++++++++++++---------------
>  hw/ide/internal.h | 21 +++++++++++++++++++++
>  hw/ide/macio.c    |  2 ++
>  4 files changed, 44 insertions(+), 22 deletions(-)"
> 
> this letter was sent at "04/06/2016 09:40 AM"
> 
> Den

Odd. Maybe because the "V4" tags were missing, it confused our patches
tool? You're right, I don't see it in the mail itself, sorry.

--js