On Thu, 7 Nov 2019 at 15:12, Alex Bennée wrote:
>
>
> > For a fix, we could put this at the beginning of qemu_set_log_filename().
> > if (logfilename) {
> > g_free(logfilename);
> > logfilename = NULL;
> > }
>
> g_free(logfilename) should be safe against NULL. However we need to
> ensure
Agree with all the suggestions below. These are great ideas. Will
make the changes.
Thanks,
-Rob Foley
On Thu, 7 Nov 2019 at 11:32, Alex Bennée wrote:
>
>
> Robert Foley writes:
>
> > One test ensures that the logfile handle is still valid even if
> > the logfile is changed during logging.
>
Robert Foley writes:
> Thanks for providing the stack trace.
>
> We debugged this and it seems to come about because of an interesting
> circumstance. We added our new tests after a pre-existing test,
> parse_path(), which runs into an issue, a dangling pointer, which
> could lead to a double
Thanks for providing the stack trace.
We debugged this and it seems to come about because of an interesting
circumstance. We added our new tests after a pre-existing test,
parse_path(), which runs into an issue, a dangling pointer, which
could lead to a double free. There were no other tests
Alex Bennée writes:
> Robert Foley writes:
>
>> One test ensures that the logfile handle is still valid even if
>> the logfile is changed during logging.
>> The other test validates that the logfile handle remains valid under
>> the logfile lock even if the logfile is closed.
Also this
Robert Foley writes:
> One test ensures that the logfile handle is still valid even if
> the logfile is changed during logging.
> The other test validates that the logfile handle remains valid under
> the logfile lock even if the logfile is closed.
>
> Signed-off-by: Robert Foley
> ---
>
One test ensures that the logfile handle is still valid even if
the logfile is changed during logging.
The other test validates that the logfile handle remains valid under
the logfile lock even if the logfile is closed.
Signed-off-by: Robert Foley
---
tests/test-logging.c | 74