4:06 a.m.
https://bugzilla.redhat.com/show_bug.cgi?id=1251886
We do not allow delete an iothread which id is 0 in
virDomainDelIOThread, but allow it in virDomainAddIOThread,
Also we will output an error when parse an iothread which id
is 0.
Add a check for iothread_id in virDomainAddIOThread to fix
it.
Signed-off-by: Luyao Huang <lhuang(a)redhat.com>
---
Is that okay to fix it in this place ?
src/libvirt-domain.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/libvirt-domain.c b/src/libvirt-domain.c
index 837933f..8506942 100644
--- a/src/libvirt-domain.c
+++ b/src/libvirt-domain.c
@@ -7987,6 +7987,7 @@ virDomainAddIOThread(virDomainPtr domain,
virCheckDomainReturn(domain, -1);
virCheckReadOnlyGoto(domain->conn->flags, error);
+ virCheckNonZeroArgGoto(iothread_id, error);
conn = domain->conn;
--
1.8.3.1
4:23 a.m.
On Mon, Aug 10, 2015 at 17:06:31 +0800, Luyao Huang wrote:
https://bugzilla.redhat.com/show_bug.cgi?id=1251886
We do not allow delete an iothread which id is 0 in
virDomainDelIOThread, but allow it in virDomainAddIOThread,
Also we will output an error when parse an iothread which id
is 0.
Add a check for iothread_id in virDomainAddIOThread to fix
it.
The limitation that iothread id shall not be 0 comes from the qemu
implementation so I think that we could possibly want to have iothread
id 0 in the future.
I think the check should be done in the qemu driver.
Peter
4:27 a.m.
On 08/10/2015 05:23 PM, Peter Krempa wrote:
On Mon, Aug 10, 2015 at 17:06:31 +0800, Luyao Huang wrote:
> https://bugzilla.redhat.com/show_bug.cgi?id=1251886
>
> We do not allow delete an iothread which id is 0 in
> virDomainDelIOThread, but allow it in virDomainAddIOThread,
> Also we will output an error when parse an iothread which id
> is 0.
>
> Add a check for iothread_id in virDomainAddIOThread to fix
> it.
The limitation that iothread id shall not be 0 comes from the qemu
implementation so I think that we could possibly want to have iothread
id 0 in the future.
I think the check should be done in the qemu driver.
Make sense, i will move the check in qemu driver, also i will change the
code for iothread pin/delete
Thanks for your quick review.
Peter
Luyao
8:17 a.m.
On 08/10/2015 05:23 AM, Peter Krempa wrote:
On Mon, Aug 10, 2015 at 17:06:31 +0800, Luyao Huang wrote:
> https://bugzilla.redhat.com/show_bug.cgi?id=1251886
>
> We do not allow delete an iothread which id is 0 in
> virDomainDelIOThread, but allow it in virDomainAddIOThread,
> Also we will output an error when parse an iothread which id
> is 0.
>
> Add a check for iothread_id in virDomainAddIOThread to fix
> it.
The limitation that iothread id shall not be 0 comes from the qemu
implementation so I think that we could possibly want to have iothread
id 0 in the future.
Sort of.. The initial implementation used 0 (zero) as a marker of sorts
for no iothreads, so 'iothread0' made no sense. At that time there was
no add/del a specific thread.
However, now that there are iothreadid's to manage the iothreads<N>
value, it perhaps makes less sense to restrict usage of 0 (zero) except
of course for historical purposes. There probably would need to be
checks in multiple places adjusted though in order to manage the
iothreadid's which 'assume' we start at 1 and count upwards when there
is no <iothread id="#"/> provided in the domain XML.
John
I think the check should be done in the qemu driver.
Peter
--
libvir-list mailing list
libvir-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
8:44 a.m.
On 08/10/2015 09:17 PM, John Ferlan wrote:
On 08/10/2015 05:23 AM, Peter Krempa wrote:
> On Mon, Aug 10, 2015 at 17:06:31 +0800, Luyao Huang wrote:
>> https://bugzilla.redhat.com/show_bug.cgi?id=1251886
>>
>> We do not allow delete an iothread which id is 0 in
>> virDomainDelIOThread, but allow it in virDomainAddIOThread,
>> Also we will output an error when parse an iothread which id
>> is 0.
>>
>> Add a check for iothread_id in virDomainAddIOThread to fix
>> it.
> The limitation that iothread id shall not be 0 comes from the qemu
> implementation so I think that we could possibly want to have iothread
> id 0 in the future.
>
Sort of.. The initial implementation used 0 (zero) as a marker of sorts
for no iothreads, so 'iothread0' made no sense. At that time there was
no add/del a specific thread.
However, now that there are iothreadid's to manage the iothreads<N>
value, it perhaps makes less sense to restrict usage of 0 (zero) except
of course for historical purposes. There probably would need to be
checks in multiple places adjusted though in order to manage the
iothreadid's which 'assume' we start at 1 and count upwards when there
is no <iothread id="#"/> provided in the domain XML.
Oh, i see, that is why i cannot find any reason to forbid iothread0 in
qemu source code, thanks a lot for clearly explanation.
Indeed, if we allow 0 as a valid iothread id and start at 0 will need
some work to fix the possible issue happen on a guest with iothread
during update, IMHO: it worth to do that :)
Thanks a lot for your reply
John
Luyao
> I think the check should be done in the qemu driver.
>
> Peter
>
>
>
> --
> libvir-list mailing list
> libvir-list(a)redhat.com
> https://www.redhat.com/mailman/listinfo/libvir-list
>
3392
days inactive
3392
days old
4 comments
4 participants
participants (4)
-
John Ferlan -
lhuang -
Luyao Huang
-
Peter Krempa