From mboxrd@z Thu Jan 1 00:00:00 1970 From: skannan@codeaurora.org (Saravana Kannan) Date: Mon, 14 Jul 2014 12:10:21 -0700 Subject: [PATCH v2] cpufreq: Don't destroy/realloc policy/sysfs on hotplug/suspend In-Reply-To: References: <1404959850-11617-1-git-send-email-skannan@codeaurora.org> <1405052287-4744-1-git-send-email-skannan@codeaurora.org> <53C0A648.9080909@codeaurora.org> Message-ID: <53C42B1D.8070301@codeaurora.org> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 07/13/2014 11:13 PM, Viresh Kumar wrote: > On 12 July 2014 08:36, Saravana Kannan wrote: >> On 07/10/2014 11:19 PM, Viresh Kumar wrote: >> >>> >>> Please make sure you take care of these issues: >>> - suspend/resume >>> - hotplug >>> - module insert/remove >> >> Ok, I was just at the current code. Does cpufreq_unregister_driver() even >> really work correctly as it stands? >> >> It doesn't even seem to stop any of the governors/policies before it just >> set the cpufreq_driver pointer to NULL. >> >> So, technically my v2 patch doesn't even make anything worse when it comes >> to unregistering the cpufreq driver. > > Are you really sure about this? I have tested this *myself* earlier.. > > subsys_interface_unregister() should take care of stopping/freeing governor > stuff.. > I was asking this question based on looking at the code. Didn't actually try it -- sent it just before being done for the day. I didn't know about the subsys_interface_unregister() coming into play here. I'll take a look. Thanks for the pointer. -Saravana -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, hosted by The Linux Foundation