Dear Oliver,
Am 10.08.25 um 10:20 schrieb Oliver Chang:
Thank you for the patch. For the summary, I’d use imperative mood and do
not add a dot/period at the end:
Fix heap-buffer-overflow in `compute_seq_size`
Thank you for the suggestion, I will make sure to do this for future
commit summaries. What do I need to do here in this instance to make
this change? Do I need to resend the patch with a new subject?
If you want to, you could send v2. (`git format-patch -v2`)
The last comment says:
This issue was migrated from crbug.com/oss-fuzz/51480?no_tracker_redirect=1
But that URL gives *Page Not Found*.
Apologies for that. We had a migration from a legacy issue tracking
system, and that legacy system has since been turned down.
Understood. No problem.
https://oss-fuzz.com/testcase-detail/5896441415729152
I am unable to access this without logging in.
The emails that can access this are the ones listed here:
https://github.com/google/oss-fuzz/blob/ef1c29822d62470fb6b0af7b73412d245d05f80c/projects/bluez/project.yaml#L3.
Are there any other emails we should add?
I am not a maintainer, so, as these might security sensitive, I guess
it’s fine, that I am unable to access it in general.
These emails also receive automatic email notifications whenever
OSS-Fuzz finds a new bug. Note though, to view the oss-fuzz.com
report, you'll need either a GitHub or Google account associated with
the email.
https://oss-fuzz.com/testcase-detail/5896441415729152 contains the
ASan stacktrace, which I've also included in my earlier email.
Understood, in the cover letter [1]. Thank you! I personally prefer
everything to be in the commit message, so it’s self-contained.
With your patch and the test case, what error is logged now?
There is still a memory leak detected by ASan that's unrelated to this
patch/issue, but the buffer overflow report is gone.
I don't see any other log messages, including the ones I added in the
patch. This is likely because `sdp_xml_parse_record` calls
`g_markup_parse_context_parse` with a NULL `error`, so any parsing
errors encountered are not propagated.
```
if (g_markup_parse_context_parse(ctx, data, size, NULL) == FALSE) {
```
It seems useful to enable propagating of parsing errors to
`sdp_xml_parse_record` in the future. But if preferred, I can remove
the logging I added since they're not going anywhere right now.
Thank you for analyzing this, and giving the reason. I’d keep it like
this, but add a note to the commit message.
Thank you again and kind regards,
Paul
[1]:
https://lore.kernel.org/all/20250810064656.1810093-2-ochang@xxxxxxxxxx/#t
