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 vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 0A6C3C07CA9 for ; Thu, 30 Nov 2023 12:36:48 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1345377AbjK3Mgj (ORCPT ); Thu, 30 Nov 2023 07:36:39 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:43942 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1345388AbjK3Mgg (ORCPT ); Thu, 30 Nov 2023 07:36:36 -0500 Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by lindbergh.monkeyblade.net (Postfix) with ESMTP id 8587C10DE; Thu, 30 Nov 2023 04:36:41 -0800 (PST) Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 6B0CD1042; Thu, 30 Nov 2023 04:37:27 -0800 (PST) Received: from pluto (unknown [172.31.20.19]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id 6AA393F5A1; Thu, 30 Nov 2023 04:36:39 -0800 (PST) Date: Thu, 30 Nov 2023 12:36:36 +0000 From: Cristian Marussi To: Sibi Sankar Cc: sudeep.holla@arm.com, linux-arm-msm@vger.kernel.org, linux-kernel@vger.kernel.org, quic_mdtipton@quicinc.com, linux-arm-kernel@lists.infradead.org, quic_asartor@quicinc.com, quic_lingutla@quicinc.com Subject: Re: [PATCH 1/3] firmware: arm_scmi: Fix null pointer dereference during fastchannel init Message-ID: References: <20231129065748.19871-1-quic_sibis@quicinc.com> <20231129065748.19871-2-quic_sibis@quicinc.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20231129065748.19871-2-quic_sibis@quicinc.com> Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Nov 29, 2023 at 12:27:46PM +0530, Sibi Sankar wrote: > The scmi_perf_domain_lookup requires the protocol handle to have the > private data set, which is yet to happen during the fastchannel init > scenario. This results in a null pointer dereference. Fix this by using > the pre-populated perf_dom_info to pass on the required information > instead. > > Call trace: > scmi_perf_protocol_init+0x434/0x678 > scmi_get_protocol_instance+0x168/0x29c > scmi_devres_protocol_instance_get+0x50/0xa0 > scmi_devm_protocol_get+0x20/0x50 > scmi_cpufreq_probe+0x34/0xd4 > scmi_dev_probe+0x28/0x3c > really_probe+0x148/0x2ac > __driver_probe_device+0x78/0x12c > driver_probe_device+0x40/0x160 > __device_attach_driver+0xb8/0x134 > bus_for_each_drv+0x80/0xdc > __device_attach+0xa8/0x1b0 > device_initial_probe+0x14/0x20 > bus_probe_device+0xa8/0xac > device_add+0x5cc/0x778 > device_register+0x20/0x30 > __scmi_device_create.part.0+0xec/0x1cc > scmi_device_create+0x180/0x1c4 > scmi_create_protocol_devices+0x4c/0xb0 > scmi_probe+0x660/0x738 > platform_probe+0x68/0xdc > really_probe+0x148/0x2ac > __driver_probe_device+0x78/0x12c > driver_probe_device+0x40/0x160 > __device_attach_driver+0xb8/0x134 > bus_for_each_drv+0x80/0xdc > __device_attach+0xa8/0x1b0 > device_initial_probe+0x14/0x20 > bus_probe_device+0xa8/0xac > deferred_probe_work_func+0x88/0xc0 > process_one_work+0x13c/0x264 > worker_thread+0x32c/0x438 > kthread+0x118/0x11c > ret_from_fork+0x10/0x20 > > Fixes: 619bc6e034f3 ("firmware: arm_scmi: Populate fastchannel info only if set operations are allowed") > Signed-off-by: Sibi Sankar > --- Yes indeed, I made this same error in the past and then I missed it when reviewing the offending patch :< ... Reviewed-by: Cristian Marussi Thanks, Cristian > drivers/firmware/arm_scmi/perf.c | 19 +++++++------------ > 1 file changed, 7 insertions(+), 12 deletions(-) > > diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c > index d1323c5d9c27..a648521e04a3 100644 > --- a/drivers/firmware/arm_scmi/perf.c > +++ b/drivers/firmware/arm_scmi/perf.c > @@ -759,40 +759,35 @@ static int scmi_perf_level_limits_notify(const struct scmi_protocol_handle *ph, > } > > static void scmi_perf_domain_init_fc(const struct scmi_protocol_handle *ph, > - u32 domain, struct scmi_fc_info **p_fc) > + struct perf_dom_info *dom) > { > struct scmi_fc_info *fc; > - struct perf_dom_info *dom; > - > - dom = scmi_perf_domain_lookup(ph, domain); > - if (IS_ERR(dom)) > - return; > > fc = devm_kcalloc(ph->dev, PERF_FC_MAX, sizeof(*fc), GFP_KERNEL); > if (!fc) > return; > > ph->hops->fastchannel_init(ph, PERF_DESCRIBE_FASTCHANNEL, > - PERF_LEVEL_GET, 4, domain, > + PERF_LEVEL_GET, 4, dom->id, > &fc[PERF_FC_LEVEL].get_addr, NULL); > > ph->hops->fastchannel_init(ph, PERF_DESCRIBE_FASTCHANNEL, > - PERF_LIMITS_GET, 8, domain, > + PERF_LIMITS_GET, 8, dom->id, > &fc[PERF_FC_LIMIT].get_addr, NULL); > > if (dom->info.set_perf) > ph->hops->fastchannel_init(ph, PERF_DESCRIBE_FASTCHANNEL, > - PERF_LEVEL_SET, 4, domain, > + PERF_LEVEL_SET, 4, dom->id, > &fc[PERF_FC_LEVEL].set_addr, > &fc[PERF_FC_LEVEL].set_db); > > if (dom->set_limits) > ph->hops->fastchannel_init(ph, PERF_DESCRIBE_FASTCHANNEL, > - PERF_LIMITS_SET, 8, domain, > + PERF_LIMITS_SET, 8, dom->id, > &fc[PERF_FC_LIMIT].set_addr, > &fc[PERF_FC_LIMIT].set_db); > > - *p_fc = fc; > + dom->fc_info = fc; > } > > static int scmi_dvfs_device_opps_add(const struct scmi_protocol_handle *ph, > @@ -1102,7 +1097,7 @@ static int scmi_perf_protocol_init(const struct scmi_protocol_handle *ph) > scmi_perf_describe_levels_get(ph, dom, version); > > if (dom->perf_fastchannels) > - scmi_perf_domain_init_fc(ph, dom->id, &dom->fc_info); > + scmi_perf_domain_init_fc(ph, dom); > } > > ret = devm_add_action_or_reset(ph->dev, scmi_perf_xa_destroy, pinfo); > -- > 2.17.1 >