From: Andrii Anisov
An IRQ assigned to guest always has an action. This removes
another odd check on guest IRQ path. Also getting an unknown
interrupt is very unlikely on a non-debug platform.
Signed-off-by: Andrii Anisov
---
xen/arch/arm/irq.c | 14 +++---
1 file changed, 7 insertions
Hi Andrii,
On 28/11/2018 21:32, Andrii Anisov wrote:
From: Andrii Anisov
An IRQ assigned to guest always has an action. This removes
another odd check on guest IRQ path.
And you can't see any potential race in that code happening in the future?
Also getting an unknown
interrupt is very unl
On 11.12.18 16:48, Julien Grall wrote:
And you can't see any potential race in that code happening in the future?
It is protected with `desc->lock` so far. If one decided to get it from under
the lock, the race is possible with `release_irq()`.
Also getting an unknown
interrupt is very unli
On 12/12/2018 11:30, Andrii Anisov wrote:
On 11.12.18 16:48, Julien Grall wrote:
And you can't see any potential race in that code happening in the future?
It is protected with `desc->lock` so far. If one decided to get it from under
the lock, the race is possible with `release_irq()`.
A
Julien,
On 12.12.18 13:59, Julien Grall wrote:
An ASSERT(...) is already embed in irq_get_guest_info(desc).
Yep.
I thought about the ASSERT(...) over the current printk yesterday. But I
discarded it because it does not give you more information if something went
really wrong as the stack t
Hi,
On 12/12/2018 13:51, Andrii Anisov wrote:
Julien,
On 12.12.18 13:59, Julien Grall wrote:
An ASSERT(...) is already embed in irq_get_guest_info(desc).
Yep.
I thought about the ASSERT(...) over the current printk yesterday. But I
discarded it because it does not give you more information
On 12.12.18 16:49, Julien Grall wrote:
ASSERT only tells you that desc->action was NULL. It does not tell you which IRQ
has the desc->action == NULL.
Ah, yes.
It is a bit a shame we don't have way to provide another message with ASSERT to
help you debugging.
We might have implemented an a
On 12/12/2018 14:58, Andrii Anisov wrote:
On 12.12.18 16:49, Julien Grall wrote:
ASSERT only tells you that desc->action was NULL. It does not tell you which
IRQ has the desc->action == NULL.
Ah, yes.
It is a bit a shame we don't have way to provide another message with ASSERT
to help you
On 12.12.18 17:08, Julien Grall wrote:
Is it just because it adds a couple more instruction in the guest case?
And add unlikely for XEN IRQ case. That was the idea.
The check is mainly there to catch error in debug build.
I supposed a race with `release_irq()`, but found out that we are safe