From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 11B11C4167B for ; Tue, 28 Nov 2023 12:38:14 +0000 (UTC) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1r7xLB-0001fx-Rp; Tue, 28 Nov 2023 07:37:25 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]) by lists.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1r7xL7-0001fZ-El for qemu-devel@nongnu.org; Tue, 28 Nov 2023 07:37:21 -0500 Received: from us-smtp-delivery-124.mimecast.com ([170.10.133.124]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1r7xL1-00041n-2H for qemu-devel@nongnu.org; Tue, 28 Nov 2023 07:37:18 -0500 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1701175030; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: in-reply-to:in-reply-to:references:references; bh=0TXXlHcEIF/S9wI+CACdibWbx6pm9OnLsFI9Pz4a9hQ=; b=H2gFVfipuQdZDQOfrQTjQ2LGom+30ovOTGnOFRxIagTTMN1c1+73lziIOxaQj18N7t0iBo VB23TbhE867LBqmQ0y5S/zjw861xuoQJzY7gQMBRroriRtlPmzQga8s8+LvKRxTsPkjWqO +R08EMigvCV32vU1kNtjohQQ2K7fnVI= Received: from mimecast-mx02.redhat.com (mimecast-mx02.redhat.com [66.187.233.88]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-342-gbsXV41BOIeJeiiW-dAKBA-1; Tue, 28 Nov 2023 07:37:05 -0500 X-MC-Unique: gbsXV41BOIeJeiiW-dAKBA-1 Received: from smtp.corp.redhat.com (int-mx03.intmail.prod.int.rdu2.redhat.com [10.11.54.3]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id A063B101A54C; Tue, 28 Nov 2023 12:37:04 +0000 (UTC) Received: from blackfin.pond.sub.org (unknown [10.39.192.148]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 37E7F1121307; Tue, 28 Nov 2023 12:37:04 +0000 (UTC) Received: by blackfin.pond.sub.org (Postfix, from userid 1000) id 4279B21E6A1F; Tue, 28 Nov 2023 13:37:02 +0100 (CET) From: Markus Armbruster To: Het Gala Cc: qemu-devel@nongnu.org, prerna.saxena@nutanix.com, quintela@redhat.com, berrange@redhat.com, peter.maydell@linaro.org, farosas@suse.de Subject: Re: [PATCH] 'channel' and 'addr' in qmp_migrate() are not auto-freed. migrate_uri_parse() allocates memory which is returned to 'channel', which is leaked because there is no code for freeing 'channel' or 'addr'. So, free addr and channel to avoid memory leak. 'addr' does shallow copying of channel->addr, hence free 'channel' itself and deep free contents of 'addr' References: <20231128062520.36456-1-het.gala@nutanix.com> <87a5qy4aag.fsf@pond.sub.org> <875y1m19ls.fsf@pond.sub.org> <7d7c582a-50fa-4f66-a4c1-0753f69fc7d8@nutanix.com> Date: Tue, 28 Nov 2023 13:37:02 +0100 In-Reply-To: <7d7c582a-50fa-4f66-a4c1-0753f69fc7d8@nutanix.com> (Het Gala's message of "Tue, 28 Nov 2023 17:05:54 +0530") Message-ID: <871qcavytt.fsf@pond.sub.org> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/28.2 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain X-Scanned-By: MIMEDefang 3.4.1 on 10.11.54.3 Received-SPF: pass client-ip=170.10.133.124; envelope-from=armbru@redhat.com; helo=us-smtp-delivery-124.mimecast.com X-Spam_score_int: -20 X-Spam_score: -2.1 X-Spam_bar: -- X-Spam_report: (-2.1 / 5.0 requ) BAYES_00=-1.9, DKIMWL_WL_HIGH=-0.001, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, RCVD_IN_DNSWL_NONE=-0.0001, RCVD_IN_MSPIKE_H3=0.001, RCVD_IN_MSPIKE_WL=0.001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001, T_SCC_BODY_TEXT_LINE=-0.01 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Het Gala writes: > On 28/11/23 3:29 pm, Markus Armbruster wrote: >> Het Gala writes: >> >>> On 28/11/23 12:46 pm, Markus Armbruster wrote: >>>> Your commit message is all in one line. You need to format it like >>>> >>>> migration: Plug memory leak >>>> >>>> 'channel' and 'addr' in qmp_migrate() are not auto-freed. >>>> migrate_uri_parse() allocates memory which is returned to 'channel', >>>> which is leaked because there is no code for freeing 'channel' or >>>> 'addr'. So, free addr and channel to avoid memory leak. 'addr' >>>> does shallow copying of channel->addr, hence free 'channel' itself >>>> and deep free contents of 'addr'. >>>> >>>> Het Gala writes: >>> Yeah, I made the changes in v2 patchset. >>>>> --- >>>>> migration/migration.c | 2 ++ >>>>> 1 file changed, 2 insertions(+) >>>>> >>>>> diff --git a/migration/migration.c b/migration/migration.c >>>>> index 28a34c9068..29efb51b62 100644 >>>>> --- a/migration/migration.c >>>>> +++ b/migration/migration.c >>>>> @@ -2004,6 +2004,8 @@ void qmp_migrate(const char *uri, bool has_channels, >>>>> MIGRATION_STATUS_FAILED); >>>>> block_cleanup_parameters(); >>>>> } >>>>> + g_free(channel); >>>>> + qapi_free_MigrationAddress(addr); >>>>> if (local_err) { >>>>> if (!resume_requested) { >>>> 2. hmp_migrate() >>>> >>>> hmp_migrate() allocates @channel with migrate_uri_parse(), adds it to >>>> list @caps, passes @caps to qmp_migrate(), then frees @caps with >>>> qapi_free_MigrationChannelList(). >>> Markus, sorry if I was not able to put point clearly, what I meant is that the local 'channel' variable used in qmp_migrate() i.e. >>> >>> 'MigrationChannel *channel = NULL', is defined in qmp_migrate() and if the user opts for 'uri' then '@channels' coming from hmp_migrate() will be NULL, and then migrate_uri_parse() will populate memory into 'channel', and that is not getting freed after it's use is over. >>> >>> I think, that is where memory leak might be happening ? >> Aha! >> >> if (uri && has_channels) { >> error_setg(errp, "'uri' and 'channels' arguments are mutually " >> "exclusive; exactly one of the two should be present in " >> "'migrate' qmp command "); >> return; >> } else if (channels) { >> /* To verify that Migrate channel list has only item */ >> if (channels->next) { >> error_setg(errp, "Channel list has more than one entries"); >> return; >> } >> channel = channels->value; >> } else if (uri) { >> /* caller uses the old URI syntax */ >> if (!migrate_uri_parse(uri, &channel, errp)) { >> return; >> } >> } else { >> error_setg(errp, "neither 'uri' or 'channels' argument are " >> "specified in 'migrate' qmp command "); >> return; >> } >> >> At this point, @channel is either channels->value, or from >> migrate_uri_parse(). >> >> We must not free in the former case, we must free in the latter case, >> >> Before your patch, we don't free. Memory leak in the latter case. >> >> Afterwards, we free. Double-free in the former case. >> >> You could guard the free, like so: >> >> if (uri) { >> qapi_free_MigrationChannel(channel); >> } > > Yeah, you explained it right. The above solution seems fine to me. > > I am just curious to ask this: can we use g_autoptr() for local vaiable 'channel' and 'addr' ? As we are not passing these variables to the caller function, nor we are trying to transfer their ownership to another variable, so use of g_steal_pointer() also might not be required ? You could try something like diff --git a/migration/migration.c b/migration/migration.c index 28a34c9068..7faa9c2ebd 100644 --- a/migration/migration.c +++ b/migration/migration.c @@ -1932,7 +1932,7 @@ void qmp_migrate(const char *uri, bool has_channels, bool resume_requested; Error *local_err = NULL; MigrationState *s = migrate_get_current(); - MigrationChannel *channel = NULL; + g_autoptr(MigrationChannel) channel = NULL; MigrationAddress *addr = NULL; /* @@ -1949,18 +1949,18 @@ void qmp_migrate(const char *uri, bool has_channels, error_setg(errp, "Channel list has more than one entries"); return; } - channel = channels->value; + addr = channels->value->addr; } else if (uri) { /* caller uses the old URI syntax */ if (!migrate_uri_parse(uri, &channel, errp)) { return; } + addr = channel->addr; } else { error_setg(errp, "neither 'uri' or 'channels' argument are " "specified in 'migrate' qmp command "); return; } - addr = channel->addr; /* transport mechanism not suitable for migration? */ if (!migration_channels_and_transport_compatible(addr, errp)) { Untested. >> >> By the way, I the conditional shown above is harder to understand than >> necessary. I like to get the errors out of the way at the beginning, >> like this: >> >> if (uri && has_channels) { >> error_setg(errp, "'uri' and 'channels' arguments are mutually " >> "exclusive; exactly one of the two should be present in " >> "'migrate' qmp command "); >> return; >> } >> if (!uri && !has_channels) { >> error_setg(errp, "neither 'uri' or 'channels' argument are " >> "specified in 'migrate' qmp command "); >> return; >> } >> >> if (channels) { >> /* To verify that Migrate channel list has only item */ >> >> Or even >> >> if (!uri == !has_channels) { >> error_setg(errp, "need either 'uri' or 'channels' argument") >> return; >> } >> >> Suggestion, not demand. If you do it, separate patch. >> > Yeah, I probably opted for 'if, else if' block because I found it easy to have all 4 options in that manner. > > '!uri == !has_channels' is same as '!uri && !has_channels' right ? No. It's "either both are null/false, or both are non-null/true". > Now looking at the Qemu code, it is better to have conditional statements the way you mentioned. Will do it in a separate patch. > > > Regards, > > Het Gala