Re: [PATCH v2 11/11] bisect: handle dup() failure when redirecting stdout
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Aug 6, 2026, 17:31 UTC
- Message-ID
- <xmqqtsp7jguc.fsf@gitster.g>
- In-Reply-To
- <20260806154103.GA1625706@coredump.intra.peff.net>
Jeff King <peff@peff.net> writes:
Show 21 quoted lines
> On Wed, Aug 05, 2026 at 06:31:00PM +0000, Johannes Schindelin via GitGitGadget wrote:
>
>> diff --git a/builtin/bisect.c b/builtin/bisect.c
>> index ceb60b0626..733d28d377 100644
>> --- a/builtin/bisect.c
>> +++ b/builtin/bisect.c
>> @@ -1308,7 +1308,12 @@ static int bisect_run(struct bisect_terms *terms, int argc, const char **argv)
>>
>> fflush(stdout);
>> saved_stdout = dup(1);
>> - dup2(temporary_stdout_fd, 1);
>> + if (saved_stdout < 0 ||
>> + dup2(temporary_stdout_fd, 1) < 0) {
>> + res = error_errno(_("could not duplicate stdout"));
>> + close(temporary_stdout_fd);
>> + break;
>> + }
>
> Ironically this produces a new Coverity complaint. ;)
>
> If dup2() fails, then we break out of the loop, leaking saved_stdout.I didn't notice it while I was looking at this part, and wondering if we can (and should) do anything if close() failed there.