From mboxrd@z Thu Jan 1 00:00:00 1970 From: Suman Anna Subject: Re: [PATCH] ARM: OMAP2+: hwmod: Fix _wait_target_ready() for hwmods without sysc Date: Tue, 13 Jan 2015 19:56:42 -0600 Message-ID: <54B5CCDA.4040304@ti.com> References: <1418917790-1791-1-git-send-email-rogerq@ti.com> <5492F84E.7020706@ti.com> <54AA4CD4.40105@ti.com> <54AB9983.1040703@ti.com> <54AC1800.1020406@ti.com> Mime-Version: 1.0 Content-Type: text/plain; charset="windows-1252" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Paul Walmsley Cc: Lokesh Vutla , Roger Quadros , tony@atomide.com, santosh.shilimkar@oracle.com, t-kristo@ti.com, nm@ti.com, nsekhar@ti.com, bcousson@baylibre.com, linux-omap@vger.kernel.org, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org List-Id: linux-omap@vger.kernel.org Hi Paul, On 01/13/2015 05:29 PM, Paul Walmsley wrote: > Hi Suman, > > thanks for pitching in on this! > > On Tue, 6 Jan 2015, Suman Anna wrote: > >> You have removed the return from the above block on failure. If any DT >> entry doesn't have the reg property, this will hang the kernel boot. >> Just remove the "reg" entry from any of the existing DT, and you will >> run into the issue, this is what 6423d6df1440 ("ARM: OMAP2+: hwmod: >> check for module address space during init") fixed. > > Seems like that's the problem that we need to track down, then. If a > hwmod has no MPU-accessible registers, it should still be possible to > init the clocks for the device, etc. ... Yes true, and I should have rephrased above statement a little better - its for modules with sysc but with no reg property to supply the base for the module's SYSCONFIG or SYSSTATUS registers. The commit 6423d6df1440 has the explanation for the hang. > >> Also, are you sure you want to turn the WARN into a pr_debug, it won't >> even show during the kernel boot log if the reg base is missing. > > No, I'm not sure :-) I guess it depends how many hwmods we'll have with > no MPU-accessible registers. We don't seem to have address ranges for the > interconnects defined; we could fix that fairly easily. The WARN_ON previously was to throw a eye-catchy print for the case where hwmods have sysc but no address space defined (is an error usually, but this is what we run into during the DT conversion of a device as the hwmod and DTS changes come in through separate topic branches). I still think that the sysc check should be before the check for _HWMOD_NO_MPU_PORT, a module with sysc mandates it has an MPU port. regards Suman From mboxrd@z Thu Jan 1 00:00:00 1970 From: s-anna@ti.com (Suman Anna) Date: Tue, 13 Jan 2015 19:56:42 -0600 Subject: [PATCH] ARM: OMAP2+: hwmod: Fix _wait_target_ready() for hwmods without sysc In-Reply-To: References: <1418917790-1791-1-git-send-email-rogerq@ti.com> <5492F84E.7020706@ti.com> <54AA4CD4.40105@ti.com> <54AB9983.1040703@ti.com> <54AC1800.1020406@ti.com> Message-ID: <54B5CCDA.4040304@ti.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi Paul, On 01/13/2015 05:29 PM, Paul Walmsley wrote: > Hi Suman, > > thanks for pitching in on this! > > On Tue, 6 Jan 2015, Suman Anna wrote: > >> You have removed the return from the above block on failure. If any DT >> entry doesn't have the reg property, this will hang the kernel boot. >> Just remove the "reg" entry from any of the existing DT, and you will >> run into the issue, this is what 6423d6df1440 ("ARM: OMAP2+: hwmod: >> check for module address space during init") fixed. > > Seems like that's the problem that we need to track down, then. If a > hwmod has no MPU-accessible registers, it should still be possible to > init the clocks for the device, etc. ... Yes true, and I should have rephrased above statement a little better - its for modules with sysc but with no reg property to supply the base for the module's SYSCONFIG or SYSSTATUS registers. The commit 6423d6df1440 has the explanation for the hang. > >> Also, are you sure you want to turn the WARN into a pr_debug, it won't >> even show during the kernel boot log if the reg base is missing. > > No, I'm not sure :-) I guess it depends how many hwmods we'll have with > no MPU-accessible registers. We don't seem to have address ranges for the > interconnects defined; we could fix that fairly easily. The WARN_ON previously was to throw a eye-catchy print for the case where hwmods have sysc but no address space defined (is an error usually, but this is what we run into during the DT conversion of a device as the hwmod and DTS changes come in through separate topic branches). I still think that the sysc check should be before the check for _HWMOD_NO_MPU_PORT, a module with sysc mandates it has an MPU port. regards Suman From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751566AbbANB5M (ORCPT ); Tue, 13 Jan 2015 20:57:12 -0500 Received: from bear.ext.ti.com ([192.94.94.41]:55822 "EHLO bear.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751008AbbANB5K (ORCPT ); Tue, 13 Jan 2015 20:57:10 -0500 Message-ID: <54B5CCDA.4040304@ti.com> Date: Tue, 13 Jan 2015 19:56:42 -0600 From: Suman Anna User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.3.0 MIME-Version: 1.0 To: Paul Walmsley CC: Lokesh Vutla , Roger Quadros , , , , , , , , , Subject: Re: [PATCH] ARM: OMAP2+: hwmod: Fix _wait_target_ready() for hwmods without sysc References: <1418917790-1791-1-git-send-email-rogerq@ti.com> <5492F84E.7020706@ti.com> <54AA4CD4.40105@ti.com> <54AB9983.1040703@ti.com> <54AC1800.1020406@ti.com> In-Reply-To: Content-Type: text/plain; charset="windows-1252" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Paul, On 01/13/2015 05:29 PM, Paul Walmsley wrote: > Hi Suman, > > thanks for pitching in on this! > > On Tue, 6 Jan 2015, Suman Anna wrote: > >> You have removed the return from the above block on failure. If any DT >> entry doesn't have the reg property, this will hang the kernel boot. >> Just remove the "reg" entry from any of the existing DT, and you will >> run into the issue, this is what 6423d6df1440 ("ARM: OMAP2+: hwmod: >> check for module address space during init") fixed. > > Seems like that's the problem that we need to track down, then. If a > hwmod has no MPU-accessible registers, it should still be possible to > init the clocks for the device, etc. ... Yes true, and I should have rephrased above statement a little better - its for modules with sysc but with no reg property to supply the base for the module's SYSCONFIG or SYSSTATUS registers. The commit 6423d6df1440 has the explanation for the hang. > >> Also, are you sure you want to turn the WARN into a pr_debug, it won't >> even show during the kernel boot log if the reg base is missing. > > No, I'm not sure :-) I guess it depends how many hwmods we'll have with > no MPU-accessible registers. We don't seem to have address ranges for the > interconnects defined; we could fix that fairly easily. The WARN_ON previously was to throw a eye-catchy print for the case where hwmods have sysc but no address space defined (is an error usually, but this is what we run into during the DT conversion of a device as the hwmod and DTS changes come in through separate topic branches). I still think that the sysc check should be before the check for _HWMOD_NO_MPU_PORT, a module with sysc mandates it has an MPU port. regards Suman