[CRIU] [PATCH] parasite/mem: unprotect VMAs if page dumping failed

Pavel Emelyanov xemul at virtuozzo.com
Tue Mar 29 05:03:10 PDT 2016


On 03/29/2016 02:42 PM, Dmitry Safonov wrote:
> On 03/29/2016 02:28 PM, Dmitry Safonov wrote:
>> On 03/29/2016 02:13 PM, Pavel Emelyanov wrote:
>>> On 03/28/2016 10:38 PM, Dmitry Safonov wrote:
>>>> Before patch parasite_dump_pages_seized tried to unprotect pages
>>>> by parasite command. But it had two issues:
>>>> 1. parasite doesn't wait any command, except PARASITE_CMD_FINI after
>>>>     it failed;
>>> But that's OK. The fini() call first unprotects vmas back, doesn't it?
>> Yes, you right - was confused by reject message and didn't check
>> fini for cleanup.
> What do you think, if we just return (ret) if __parasite_dump_pages_seized
> failed? (as fini will cleanup anyway)

Sounds OK.

BTW, let's introduce a fault-injection test for this case :)

-- Pavel



More information about the CRIU mailing list