Message ID | 20211111160412.11980-1-vincent.whitchurch@axis.com |
---|---|
Headers | show |
Series | virtio-i2c: Fix buffer handling | expand |
On Thu, Nov 11, 2021 at 05:04:10PM +0100, Vincent Whitchurch wrote: > This fixes a couple of bugs in the buffer handling in virtio-i2c which can > result in incorrect data on the I2C bus or memory corruption in the guest. > > I tested this on UML (virtio-uml needs a bug fix too, I will sent that out > later) with the device implementation in rust-vmm/vhost-device. > > Changes in v2: > - Added Acked-by and Fixes tags What are the plans for this patchset? > Vincent Whitchurch (2): > i2c: virtio: disable timeout handling > i2c: virtio: fix completion handling > > drivers/i2c/busses/i2c-virtio.c | 46 ++++++++++++++------------------- > 1 file changed, 19 insertions(+), 27 deletions(-) > > -- > 2.28.0
On 24-11-21, 18:55, Michael S. Tsirkin wrote: > On Thu, Nov 11, 2021 at 05:04:10PM +0100, Vincent Whitchurch wrote: > > This fixes a couple of bugs in the buffer handling in virtio-i2c which can > > result in incorrect data on the I2C bus or memory corruption in the guest. > > > > I tested this on UML (virtio-uml needs a bug fix too, I will sent that out > > later) with the device implementation in rust-vmm/vhost-device. > > > > Changes in v2: > > - Added Acked-by and Fixes tags > > > What are the plans for this patchset? Wolfram applied the first patch, but not the second. Wolfram, you can apply that one as well, it looks okay.
> Wolfram, you can apply that one as well, it looks okay.
Is it? I read that the code only waits for the last request while
Michael suggested to wait for all of them? And he did not ack patch 2
while he acked patch 1. Did I misunderstand?
On 25-11-21, 07:24, Wolfram Sang wrote: > > > Wolfram, you can apply that one as well, it looks okay. > > Is it? I read that the code only waits for the last request while > Michael suggested to wait for all of them? And he did not ack patch 2 > while he acked patch 1. Did I misunderstand? Okay, I misread it then. To clarify, we should initialize the completion for each buffer and wait for all of them to be completed before returning back to the user. Lets wait for an update by Vincent for that then.