On Tue, Jul 2, 2019 at 10:04 AM Erik Skultety <eskultet@redhat.com> wrote:
On Thu, Jun 20, 2019 at 05:11:57PM +0200, Ilias Stamatis wrote:
The current implementation has the following bugs:
- the vm variable is accessed after calling virDomainObjEndAPI on it
- if VIR_STRDUP fails and we jump to the cleanup section, we're calling VIR_FREE on pointers for which we haven't allocated memory
- the error type VIR_DOMAIN_DISK_ERROR_NONE is used which contradicts the documentation of the API that says that disks with no errors are not reported
This patch fixes all of them and additionally reports errors only for every second disk (instead of reporting errors for all disks) which was the initial intention.
I believe each patch should focus ideally on 1 thing only. So this patch should IMO be split into 3 (doesn't matter in what order), each fixing a different bug.
Erik
Hmm. I think it's impossible to solve the first 2 separately. Because by solving the 2nd one the 1st is solved as well. The 3rd one is really adding a "-1" and a "+1" in the VIR_DOMAIN_DISK_ERROR_LAST line. But this line is touched anyway by the previous fix as well. So I thought it doesn't worth it to introduce a separate patch for it. I think the patch in general is quite small so it shouldn't bet a huge problem, but if you insist I can separate the 1-line change for fix no3. from the ones before. Ilias