On Thu, Mar 27, 2025 at 02:14:25PM +0100, Kevin Wolf wrote: > Am 26.03.2025 um 18:46 hat Stefan Hajnoczi geschrieben: > > On Wed, Mar 26, 2025 at 06:13:44PM +0100, Kevin Wolf wrote: > > > Am 25.03.2025 um 21:49 hat ~h0lyalg0rithm geschrieben: > > > > From: Suraj Shirvankar <surajshirvan...@gmail.com> > > > > > > > > Signed-off-by: Suraj Shirvankar <surajshirvan...@gmail.com> > > > > --- > > > > util/fdmon-io_uring.c | 8 +++++++- > > > > 1 file changed, 7 insertions(+), 1 deletion(-) > > > > > > As Stefan already mentioned, the commit message should say why we want > > > to set this flag and why it is correct to do so. > > > > > > Is there a reason why you change the io_uring_queue_init() call in > > > util/fdmon-io_uring.c, but not the one in block/io_uring.c? > > > > I only asked Suraj to look at util/fdmon-io_uring.c because I expect > > block/io_uring.c's io_uring context to go away soon. > > > > In my local io_uring branches I have prepared commits that replace the > > io_uring context in block/io_uring.c with aio_add_sqe() calls that use > > the AioContext's fdmon-io_uring.c io_uring context. > > Then we should either document this intention in the commit message or > make this one Based-on your changes.
Hi Suraj, Please rebase your patch on my branch here: https://gitlab.com/stefanha/qemu/-/tree/aio_add_sqe My series removes the io_uring context from block/io_uring.c, unifying it with util/fdmon-io_uring.c. That way there's no need to duplicate io_uring context setup and your SINGLE_ISSUER change only needs to be done in util/fdmon-io_uring.c. The email thread for my series is here: https://lore.kernel.org/qemu-devel/20250401142721.280287-1-stefa...@redhat.com/T/#t You can add "Based-on: 20250401142721.280287-1-stefa...@redhat.com" to the cover letter of your patch to indicate that this does not apply to qemu.git/master but on top of my patch series. Stefan
signature.asc
Description: PGP signature