[v2,0/3] Convert sparc devices to new ptimer API

Message ID 20191021134357.14266-1-peter.maydell@linaro.org
Headers show
Series
  • Convert sparc devices to new ptimer API
Related show

Message

Peter Maydell Oct. 21, 2019, 1:43 p.m.
This patchset converts the devices used by sparc machines to the new
ptimer API.

Currently the ptimer design uses a QEMU bottom-half as its mechanism
for calling back into the device model using the ptimer when the
timer has expired.  Unfortunately this design is fatally flawed,
because it means that there is a lag between the ptimer updating its
own state and the device callback function updating device state, and
guest accesses to device registers between the two can return
inconsistent device state. This was reported as a bug in a specific
timer device but it's a problem with the generic ptimer code:
https://bugs.launchpad.net/qemu/+bug/1777777

The updates to the individual ptimer devices are straightforward:
we need to add begin/commit calls around the various places that
modify the ptimer state, and use the new ptimer_init() function
to create the timer.

Changes v1->v2:
 * patches 2 and 3 are the old 1 and 2 and have been reviewed
 * patch 1 is new and removes a pointless NULL check; without
   this we'd probably have got Coverity errors when patch 3
   added a use of t->timer before the check for it being NULL

thanks
--PMM


MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

*** BLURB HERE ***

Peter Maydell (3):
  hw/timer/slavio_timer: Remove useless check for NULL t->timer
  hw/timer/grlib_gptimer.c: Switch to transaction-based ptimer API
  hw/timer/slavio_timer.c: Switch to transaction-based ptimer API

 hw/timer/grlib_gptimer.c | 28 ++++++++++++++++++++++++----
 hw/timer/slavio_timer.c  | 32 +++++++++++++++++++++-----------
 2 files changed, 45 insertions(+), 15 deletions(-)

-- 
2.20.1

Comments

Philippe Mathieu-Daudé Oct. 21, 2019, 2:06 p.m. | #1
On 10/21/19 3:43 PM, Peter Maydell wrote:
> This patchset converts the devices used by sparc machines to the new

> ptimer API.

> 

> Currently the ptimer design uses a QEMU bottom-half as its mechanism

> for calling back into the device model using the ptimer when the

> timer has expired.  Unfortunately this design is fatally flawed,

> because it means that there is a lag between the ptimer updating its

> own state and the device callback function updating device state, and

> guest accesses to device registers between the two can return

> inconsistent device state. This was reported as a bug in a specific

> timer device but it's a problem with the generic ptimer code:

> https://bugs.launchpad.net/qemu/+bug/1777777

> 

> The updates to the individual ptimer devices are straightforward:

> we need to add begin/commit calls around the various places that

> modify the ptimer state, and use the new ptimer_init() function

> to create the timer.

> 

> Changes v1->v2:

>   * patches 2 and 3 are the old 1 and 2 and have been reviewed

>   * patch 1 is new and removes a pointless NULL check; without

>     this we'd probably have got Coverity errors when patch 3

>     added a use of t->timer before the check for it being NULL

> 

> thanks

> --PMM

> 

> 

> MIME-Version: 1.0

> Content-Type: text/plain; charset=UTF-8

> Content-Transfer-Encoding: 8bit

> 

> *** BLURB HERE ***

> 

> Peter Maydell (3):

>    hw/timer/slavio_timer: Remove useless check for NULL t->timer

>    hw/timer/grlib_gptimer.c: Switch to transaction-based ptimer API

>    hw/timer/slavio_timer.c: Switch to transaction-based ptimer API


Nitpicking, maybe reorder the grlib_gptimer patch last:

   hw/timer/slavio_timer: Remove useless check for NULL t->timer
   hw/timer/slavio_timer.c: Switch to transaction-based ptimer API
   hw/timer/grlib_gptimer.c: Switch to transaction-based ptimer API
Peter Maydell Oct. 24, 2019, 12:19 p.m. | #2
On Mon, 21 Oct 2019 at 14:43, Peter Maydell <peter.maydell@linaro.org> wrote:
>

> This patchset converts the devices used by sparc machines to the new

> ptimer API.

>

> Currently the ptimer design uses a QEMU bottom-half as its mechanism

> for calling back into the device model using the ptimer when the

> timer has expired.  Unfortunately this design is fatally flawed,

> because it means that there is a lag between the ptimer updating its

> own state and the device callback function updating device state, and

> guest accesses to device registers between the two can return

> inconsistent device state. This was reported as a bug in a specific

> timer device but it's a problem with the generic ptimer code:

> https://bugs.launchpad.net/qemu/+bug/1777777

>

> The updates to the individual ptimer devices are straightforward:

> we need to add begin/commit calls around the various places that

> modify the ptimer state, and use the new ptimer_init() function

> to create the timer.

>

> Changes v1->v2:

>  * patches 2 and 3 are the old 1 and 2 and have been reviewed

>  * patch 1 is new and removes a pointless NULL check; without

>    this we'd probably have got Coverity errors when patch 3

>    added a use of t->timer before the check for it being NULL


I'm going to apply these to target-arm.next; I know they haven't
been on list long but the change since v1 is only minor and
they've all been reviewed.

thanks
-- PMM
Mark Cave-Ayland Oct. 24, 2019, 6:04 p.m. | #3
On 24/10/2019 13:19, Peter Maydell wrote:

> On Mon, 21 Oct 2019 at 14:43, Peter Maydell <peter.maydell@linaro.org> wrote:

>>

>> This patchset converts the devices used by sparc machines to the new

>> ptimer API.

>>

>> Currently the ptimer design uses a QEMU bottom-half as its mechanism

>> for calling back into the device model using the ptimer when the

>> timer has expired.  Unfortunately this design is fatally flawed,

>> because it means that there is a lag between the ptimer updating its

>> own state and the device callback function updating device state, and

>> guest accesses to device registers between the two can return

>> inconsistent device state. This was reported as a bug in a specific

>> timer device but it's a problem with the generic ptimer code:

>> https://bugs.launchpad.net/qemu/+bug/1777777

>>

>> The updates to the individual ptimer devices are straightforward:

>> we need to add begin/commit calls around the various places that

>> modify the ptimer state, and use the new ptimer_init() function

>> to create the timer.

>>

>> Changes v1->v2:

>>  * patches 2 and 3 are the old 1 and 2 and have been reviewed

>>  * patch 1 is new and removes a pointless NULL check; without

>>    this we'd probably have got Coverity errors when patch 3

>>    added a use of t->timer before the check for it being NULL

> 

> I'm going to apply these to target-arm.next; I know they haven't

> been on list long but the change since v1 is only minor and

> they've all been reviewed.


Thanks Peter! Not sure if you saw my Tested-by tag last week for the slavio (sun4m)
parts, but there were no obvious regressions that I could see under qemu-system-sparc.


ATB,

Mark.
Philippe Mathieu-Daudé Oct. 24, 2019, 6:17 p.m. | #4
On 10/24/19 8:04 PM, Mark Cave-Ayland wrote:
> On 24/10/2019 13:19, Peter Maydell wrote:

> 

>> On Mon, 21 Oct 2019 at 14:43, Peter Maydell <peter.maydell@linaro.org> wrote:

>>>

>>> This patchset converts the devices used by sparc machines to the new

>>> ptimer API.

>>>

>>> Currently the ptimer design uses a QEMU bottom-half as its mechanism

>>> for calling back into the device model using the ptimer when the

>>> timer has expired.  Unfortunately this design is fatally flawed,

>>> because it means that there is a lag between the ptimer updating its

>>> own state and the device callback function updating device state, and

>>> guest accesses to device registers between the two can return

>>> inconsistent device state. This was reported as a bug in a specific

>>> timer device but it's a problem with the generic ptimer code:

>>> https://bugs.launchpad.net/qemu/+bug/1777777

>>>

>>> The updates to the individual ptimer devices are straightforward:

>>> we need to add begin/commit calls around the various places that

>>> modify the ptimer state, and use the new ptimer_init() function

>>> to create the timer.

>>>

>>> Changes v1->v2:

>>>   * patches 2 and 3 are the old 1 and 2 and have been reviewed

>>>   * patch 1 is new and removes a pointless NULL check; without

>>>     this we'd probably have got Coverity errors when patch 3

>>>     added a use of t->timer before the check for it being NULL

>>

>> I'm going to apply these to target-arm.next; I know they haven't

>> been on list long but the change since v1 is only minor and

>> they've all been reviewed.

> 

> Thanks Peter! Not sure if you saw my Tested-by tag last week for the slavio (sun4m)

> parts, but there were no obvious regressions that I could see under qemu-system-sparc.


This was on v1:
https://www.mail-archive.com/qemu-devel@nongnu.org/msg653861.html
Peter Maydell Oct. 25, 2019, 7:32 a.m. | #5
On Thu, 24 Oct 2019 at 19:10, Mark Cave-Ayland
<mark.cave-ayland@ilande.co.uk> wrote:
>

> On 24/10/2019 13:19, Peter Maydell wrote:

> > I'm going to apply these to target-arm.next; I know they haven't

> > been on list long but the change since v1 is only minor and

> > they've all been reviewed.

>

> Thanks Peter! Not sure if you saw my Tested-by tag last week for the slavio (sun4m)

> parts, but there were no obvious regressions that I could see under qemu-system-sparc.


Yeah, I saw that, thanks for the testing. I decided that since
I'd added patch 1 I didn't quite feel comfortable carrying the
tested-by tag across.

thanks
-- PMM