| Message ID | 20180511081601.14610.39946.stgit@pasha-VirtualBox |
|---|---|
| State | New |
| Headers | show |
| Series | ps2: prevent changing irq state on save and load | expand |
On 11/05/2018 10:16, Pavel Dovgalyuk wrote: > Commit 2858ab09e6f708e381fc1a1cc87e747a690c4884 changed > PS/2 keyboard/mouse buffers to the standard size. However, its state > may change when migrating from the old buffer size and therefore irq needs > updating. But this change made wrong, because it throws the whole queue > if there are too much data instead of cropping it. > > That commit also updates irq (because the queue state may change). > But updating the irq may change the VM state (and determinism of > the execution). E.g., when replaying the execution, one may save > the VM state and the state of the interrupt controller will be updated > at the moment of saving, instead of using the recorded update events. > > This patch makes the queue update deterministic: it removes the update_irq > call and crops the queue to prevent losing the characters and changing > the required irq status. > > Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru> > --- > hw/input/ps2.c | 8 ++++++-- > 1 file changed, 6 insertions(+), 2 deletions(-) > > diff --git a/hw/input/ps2.c b/hw/input/ps2.c > index 06f5d2a..8b1931b 100644 > --- a/hw/input/ps2.c > +++ b/hw/input/ps2.c > @@ -837,7 +837,12 @@ static void ps2_common_post_load(PS2State *s) > uint8_t tmp_data[PS2_QUEUE_SIZE]; > > /* set the useful data buffer queue size, < PS2_QUEUE_SIZE */ > - size = (q->count < 0 || q->count > PS2_QUEUE_SIZE) ? 0 : q->count; > + size = q->count; > + if (q->count < 0) { > + size = 0; > + } else if (q->count > PS2_QUEUE_SIZE) { > + size = PS2_QUEUE_SIZE; > + } > > /* move the queue elements to the start of data array */ > for (i = 0; i < size; i++) { > @@ -852,7 +857,6 @@ static void ps2_common_post_load(PS2State *s) > q->rptr = 0; > q->wptr = size; > q->count = size; > - s->update_irq(s->update_arg, q->count != 0); > } > > static void ps2_kbd_reset(void *opaque) > Queued, thanks. Paolo
Paolo, have you forgot about this? Pavel Dovgalyuk > -----Original Message----- > From: Paolo Bonzini [mailto:pbonzini@redhat.com] > Sent: Friday, May 11, 2018 12:14 PM > To: Pavel Dovgalyuk; qemu-devel@nongnu.org > Cc: arei.gonglei@huawei.com; mst@redhat.com; ciro.santilli@gmail.com; > maria.klimushenkova@ispras.ru; dovgaluk@ispras.ru; kraxel@redhat.com > Subject: Re: [PATCH] ps2: prevent changing irq state on save and load > > On 11/05/2018 10:16, Pavel Dovgalyuk wrote: > > Commit 2858ab09e6f708e381fc1a1cc87e747a690c4884 changed > > PS/2 keyboard/mouse buffers to the standard size. However, its state > > may change when migrating from the old buffer size and therefore irq needs > > updating. But this change made wrong, because it throws the whole queue > > if there are too much data instead of cropping it. > > > > That commit also updates irq (because the queue state may change). > > But updating the irq may change the VM state (and determinism of > > the execution). E.g., when replaying the execution, one may save > > the VM state and the state of the interrupt controller will be updated > > at the moment of saving, instead of using the recorded update events. > > > > This patch makes the queue update deterministic: it removes the update_irq > > call and crops the queue to prevent losing the characters and changing > > the required irq status. > > > > Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru> > > --- > > hw/input/ps2.c | 8 ++++++-- > > 1 file changed, 6 insertions(+), 2 deletions(-) > > > > diff --git a/hw/input/ps2.c b/hw/input/ps2.c > > index 06f5d2a..8b1931b 100644 > > --- a/hw/input/ps2.c > > +++ b/hw/input/ps2.c > > @@ -837,7 +837,12 @@ static void ps2_common_post_load(PS2State *s) > > uint8_t tmp_data[PS2_QUEUE_SIZE]; > > > > /* set the useful data buffer queue size, < PS2_QUEUE_SIZE */ > > - size = (q->count < 0 || q->count > PS2_QUEUE_SIZE) ? 0 : q->count; > > + size = q->count; > > + if (q->count < 0) { > > + size = 0; > > + } else if (q->count > PS2_QUEUE_SIZE) { > > + size = PS2_QUEUE_SIZE; > > + } > > > > /* move the queue elements to the start of data array */ > > for (i = 0; i < size; i++) { > > @@ -852,7 +857,6 @@ static void ps2_common_post_load(PS2State *s) > > q->rptr = 0; > > q->wptr = size; > > q->count = size; > > - s->update_irq(s->update_arg, q->count != 0); > > } > > > > static void ps2_kbd_reset(void *opaque) > > > > Queued, thanks. > > Paolo
On 11/09/2018 12:18, Pavel Dovgalyuk wrote: > Paolo, have you forgot about this? > > Pavel Dovgalyuk Yes, though Gerd might be a better match for the patch. Paolo
On Tue, Sep 11, 2018 at 01:10:56PM +0200, Paolo Bonzini wrote: > On 11/09/2018 12:18, Pavel Dovgalyuk wrote: > > Paolo, have you forgot about this? > > > > Pavel Dovgalyuk > > Yes, though Gerd might be a better match for the patch. Dropped it from my patch mailbox after seeing your 'queued' message. Pavel, can you resend? thanks, Gerd
On 11/09/2018 13:25, Gerd Hoffmann wrote: > On Tue, Sep 11, 2018 at 01:10:56PM +0200, Paolo Bonzini wrote: >> On 11/09/2018 12:18, Pavel Dovgalyuk wrote: >>> Paolo, have you forgot about this? >>> >>> Pavel Dovgalyuk >> >> Yes, though Gerd might be a better match for the patch. > > Dropped it from my patch mailbox after seeing your 'queued' message. > > Pavel, can you resend? No need, I'll queue it for real now. :) Paolo
> From: Paolo Bonzini [mailto:pbonzini@redhat.com] > On 11/09/2018 13:25, Gerd Hoffmann wrote: > > On Tue, Sep 11, 2018 at 01:10:56PM +0200, Paolo Bonzini wrote: > >> On 11/09/2018 12:18, Pavel Dovgalyuk wrote: > >>> Paolo, have you forgot about this? > >>> > >>> Pavel Dovgalyuk > >> > >> Yes, though Gerd might be a better match for the patch. > > > > Dropped it from my patch mailbox after seeing your 'queued' message. > > > > Pavel, can you resend? > > No need, I'll queue it for real now. :) Thank you. Pavel Dovgalyuk
diff --git a/hw/input/ps2.c b/hw/input/ps2.c index 06f5d2a..8b1931b 100644 --- a/hw/input/ps2.c +++ b/hw/input/ps2.c @@ -837,7 +837,12 @@ static void ps2_common_post_load(PS2State *s) uint8_t tmp_data[PS2_QUEUE_SIZE]; /* set the useful data buffer queue size, < PS2_QUEUE_SIZE */ - size = (q->count < 0 || q->count > PS2_QUEUE_SIZE) ? 0 : q->count; + size = q->count; + if (q->count < 0) { + size = 0; + } else if (q->count > PS2_QUEUE_SIZE) { + size = PS2_QUEUE_SIZE; + } /* move the queue elements to the start of data array */ for (i = 0; i < size; i++) { @@ -852,7 +857,6 @@ static void ps2_common_post_load(PS2State *s) q->rptr = 0; q->wptr = size; q->count = size; - s->update_irq(s->update_arg, q->count != 0); } static void ps2_kbd_reset(void *opaque)
Commit 2858ab09e6f708e381fc1a1cc87e747a690c4884 changed PS/2 keyboard/mouse buffers to the standard size. However, its state may change when migrating from the old buffer size and therefore irq needs updating. But this change made wrong, because it throws the whole queue if there are too much data instead of cropping it. That commit also updates irq (because the queue state may change). But updating the irq may change the VM state (and determinism of the execution). E.g., when replaying the execution, one may save the VM state and the state of the interrupt controller will be updated at the moment of saving, instead of using the recorded update events. This patch makes the queue update deterministic: it removes the update_irq call and crops the queue to prevent losing the characters and changing the required irq status. Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru> --- hw/input/ps2.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)