From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51655) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VC4Z9-0005eA-H4 for qemu-devel@nongnu.org; Wed, 21 Aug 2013 05:19:08 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VC4Z0-0004f6-Il for qemu-devel@nongnu.org; Wed, 21 Aug 2013 05:18:59 -0400 Received: from e23smtp08.au.ibm.com ([202.81.31.141]:45097) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VC4Yz-0004df-CV for qemu-devel@nongnu.org; Wed, 21 Aug 2013 05:18:50 -0400 Received: from /spool/local by e23smtp08.au.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Wed, 21 Aug 2013 19:15:30 +1000 Received: from d23relay04.au.ibm.com (d23relay04.au.ibm.com [9.190.234.120]) by d23dlp03.au.ibm.com (Postfix) with ESMTP id 925CC357804E for ; Wed, 21 Aug 2013 19:18:42 +1000 (EST) Received: from d23av01.au.ibm.com (d23av01.au.ibm.com [9.190.234.96]) by d23relay04.au.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id r7L92dlK63111180 for ; Wed, 21 Aug 2013 19:02:45 +1000 Received: from d23av01.au.ibm.com (localhost [127.0.0.1]) by d23av01.au.ibm.com (8.14.4/8.14.4/NCO v10.0 AVout) with ESMTP id r7L9IaDn007323 for ; Wed, 21 Aug 2013 19:18:36 +1000 Message-ID: <52148590.8000109@linux.vnet.ibm.com> Date: Wed, 21 Aug 2013 17:17:04 +0800 From: Wenchao Xia MIME-Version: 1.0 References: <1374808842-11051-1-git-send-email-xiawenc@linux.vnet.ibm.com> <20130730120311.7f1e44a7@redhat.com> <20130820100442.1abc070d@redhat.com> In-Reply-To: <20130820100442.1abc070d@redhat.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Subject: Re: [Qemu-devel] [PATCH V8 00/13] monitor: support sub command group in auto completion and help List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Luiz Capitulino Cc: pbonzini@redhat.com, armbru@redhat.com, qemu-devel@nongnu.org 于 2013-8-20 22:04, Luiz Capitulino 写道: > On Tue, 30 Jul 2013 12:03:11 -0400 > Luiz Capitulino wrote: > >> On Fri, 26 Jul 2013 11:20:29 +0800 >> Wenchao Xia wrote: >> >>> This series make auto completion and help functions works normal for sub >>> command, by using reentrant functions. In order to do that, global variables >>> are not directly used in those functions any more. With this series, cmd_table >>> is a member of structure Monitor so it is possible to create a monitor with >>> different command table now, auto completion will work in that monitor. In >>> short, "info" is not treated as a special case now, this series ensure help >>> and auto complete function works normal for any sub command added in the future. >>> >>> Patch 5 replaced cur_mon with rs->mon, it is safe because: >>> monitor_init() calls readline_init() which initialize mon->rs, result is >>> mon->rs->mon == mon. Then qemu_chr_add_handlers() is called, which make >>> monitor_read() function take *mon as its opaque. Later, when user input, >>> monitor_read() is called, where cur_mon is set to *mon by "cur_mon = opaque". >>> If qemu's monitors run in one thread, then later in readline_handle_byte() >>> and readline_comletion(), cur_mon is actually equal to rs->mon, in another >>> word, it points to the monitor instance, so it is safe to replace *cur_mon >>> in those functions. >> >> I've applied this to qmp-next with the change I suggested for >> patch 09/13. > > Unfortunately this series brakes make check: > > GTESTER check-qtest-x86_64 > Broken pipe > GTester: last random seed: R02S3492bd34f44dd17460851643383be44d > main-loop: WARNING: I/O thread spun for 1000 iterations > make: *** [check-qtest-x86_64] Error 1 > > I debugged it (with some help from Laszlo) and the problem is that it > broke the human-monitor-command command. Any usage of this command > triggers the bug like: > > { "execute": "human-monitor-command", > "arguments": { "command-line": "info registers" } } > > It seems simple to fix, I think you just have to initialize > mon->cmd_table in qmp_human_monitor_command(), but I'd recommend two > things: > > 1. It's better to split off some/all QMP initialization from > monitor_init() and call it from qmp_human_monitor_command() > > 2. Can you please take the opportunity and test all commands using > cur_mon? Just grep for it > > Sorry for noticing this only now, but I only run make check before > sending a pull request (although this very likely shows you didn't > run it either). > My bad that not ran make check before, will fix and retry, sorry for the trouble. -- Best Regards Wenchao Xia