Juan Quintela writes:
> Peter Xu wrote:
>> On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
>>> We cannot operate on the multifd semaphores outside of the multifd
>>> channel thread
>>> because multifd_save_cleanup() can run in parallel and
>>> attempt to destroy the mutexes, which
Fabiano Rosas wrote:
> Fabiano Rosas writes:
>
>> Peter Xu writes:
>>
>>> On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
>> I think we historically stumbled upon the fact that qemu_thread_join()
>> is not the same as pthread_join(). The former takes a pointer and is not
>> safe
Fabiano Rosas wrote:
> Peter Xu writes:
>
>> On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
>>> We cannot operate on the multifd semaphores outside of the multifd
>>> channel thread
>>> because multifd_save_cleanup() can run in parallel and
>>> attempt to destroy the mutexes, whic
Peter Xu wrote:
> On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
>> We cannot operate on the multifd semaphores outside of the multifd
>> channel thread
>> because multifd_save_cleanup() can run in parallel and
>> attempt to destroy the mutexes, which causes an assert.
>>
>> Looki
On Mon, Nov 13, 2023 at 10:50:39PM -0300, Fabiano Rosas wrote:
> Peter Xu writes:
>
> > On Fri, Nov 10, 2023 at 09:05:41AM -0300, Fabiano Rosas wrote:
> >
> > [...]
> >
> >> > Then assuming we have a clear model with all these threads issue fixed
> >> > (no
> >> > matter whether we'd shrink 2N t
Peter Xu writes:
> On Fri, Nov 10, 2023 at 09:05:41AM -0300, Fabiano Rosas wrote:
>
> [...]
>
>> > Then assuming we have a clear model with all these threads issue fixed (no
>> > matter whether we'd shrink 2N threads into N threads), then what we need to
>> > do, IMHO, is making sure to join() al
On Fri, Nov 10, 2023 at 09:05:41AM -0300, Fabiano Rosas wrote:
[...]
> > Then assuming we have a clear model with all these threads issue fixed (no
> > matter whether we'd shrink 2N threads into N threads), then what we need to
> > do, IMHO, is making sure to join() all of them before destroying
Fabiano Rosas writes:
> Peter Xu writes:
>
>> On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
>>> We cannot operate on the multifd semaphores outside of the multifd
>>> channel thread
>>> because multifd_save_cleanup() can run in parallel and
>>> attempt to destroy the mutexes, wh
Peter Xu writes:
> On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
>> We cannot operate on the multifd semaphores outside of the multifd
>> channel thread
>> because multifd_save_cleanup() can run in parallel and
>> attempt to destroy the mutexes, which causes an assert.
>>
>> Loo
On Thu, Nov 09, 2023 at 01:58:56PM -0300, Fabiano Rosas wrote:
> We cannot operate on the multifd semaphores outside of the multifd
> channel thread
> because multifd_save_cleanup() can run in parallel and
> attempt to destroy the mutexes, which causes an assert.
>
> Looking at the places where we
We cannot operate on the multifd semaphores outside of the multifd
channel thread because multifd_save_cleanup() can run in parallel and
attempt to destroy the mutexes, which causes an assert.
Looking at the places where we use the semaphores aside from the
migration thread, there's only the TLS h
11 matches
Mail list logo
