Received: from eggs.gnu.org ([209.51.188.92]:38996) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1hG1Lt-0006Su-Pa for qemu-devel@nongnu.org; Mon, 15 Apr 2019 09:08:50 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1hG1Ls-0001Vh-Nk for qemu-devel@nongnu.org; Mon, 15 Apr 2019 09:08:49 -0400 Received: from mx1.redhat.com ([209.132.183.28]:42134) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1hG1Ls-0001V0-DJ for qemu-devel@nongnu.org; Mon, 15 Apr 2019 09:08:48 -0400 From: Markus Armbruster References: <20190413160732.55505-1-vsementsov@virtuozzo.com> <87zhorbytu.fsf@dusky.pond.sub.org> <3381f703-4fc1-ac3c-658c-f7399403acf9@virtuozzo.com> Date: Mon, 15 Apr 2019 15:08:44 +0200 In-Reply-To: <3381f703-4fc1-ac3c-658c-f7399403acf9@virtuozzo.com> (Vladimir Sementsov-Ogievskiy's message of "Mon, 15 Apr 2019 10:08:37 +0000") Message-ID: <874l6za00z.fsf@dusky.pond.sub.org> MIME-Version: 1.0 Content-Type: text/plain Subject: Re: [Qemu-devel] [PATCH] util/error: do not free error on error_abort List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Vladimir Sementsov-Ogievskiy Cc: "qemu-devel@nongnu.org" Vladimir Sementsov-Ogievskiy writes: > 15.04.2019 8:51, Markus Armbruster wrote: >> Vladimir Sementsov-Ogievskiy writes: >> >>> It would be nice to have Error object not freed away when debugging a >>> coredump. >>> >>> Signed-off-by: Vladimir Sementsov-Ogievskiy >>> --- >>> util/error.c | 8 +++++--- >>> 1 file changed, 5 insertions(+), 3 deletions(-) >>> >>> diff --git a/util/error.c b/util/error.c >>> index 934a78e1b1..f9180c0f30 100644 >>> --- a/util/error.c >>> +++ b/util/error.c >>> @@ -32,9 +32,11 @@ Error *error_fatal; >>> static void error_handle_fatal(Error **errp, Error *err) >>> { >>> if (errp == &error_abort) { >>> - fprintf(stderr, "Unexpected error in %s() at %s:%d:\n", >>> - err->func, err->src, err->line); >>> - error_report_err(err); >>> + error_report("Unexpected error in %s() at %s:%d: %s", >>> + err->func, err->src, err->line, error_get_pretty(err)); >>> + if (err->hint) { >>> + error_printf_unless_qmp("%s", err->hint->str); >>> + } >>> abort(); >>> } >>> if (errp == &error_fatal) { >> >> No objections to not freeing the error object on the path to abort(). >> >> You also format the error message differently. Commit 1e9b65bb1ba's >> example >> >> Unexpected error in parse_block_error_action() at .../qemu/blockdev.c:322: >> qemu-system-x86_64: -drive if=none,werror=foo: 'foo' invalid write error action >> Aborted (core dumped) >> >> changes to (guesswork, not tested): >> >> qemu-system-x86_64: -drive if=none,werror=foo: Unexpected error in parse_block_error_action() at .../qemu/blockdev.c:322: 'foo' invalid write error action >> Aborted (core dumped) > > hm. checkpatch don't allow to put \n into error_report. So, should I call error_report twice? > > Transition from fprintf to error_report is OK for you? Let's simply not mess with the formatting at all. Something like: (1) Inline error_report_err(), delete the error_free() (2) Optional: replace fprintf() by error_printf() (3) Optional: clean up the additional copy of if (err->hint) { error_printf_unless_qmp("%s", err->hint->str); } (3a) Either create a helper function, replacing all three copies, (3b) Or simply delete the new copy from the path leading to abort(), since the hint is unlikely to be useful there.