From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Martin K. Petersen" Subject: Re: [PATCH] sd: preserve sysfs updates to max_sectors_kb Date: Tue, 29 Aug 2017 21:24:12 -0400 Message-ID: References: <150309002518.8999.15900049133748830764.stgit@brunhilda> <1503090353.2622.15.camel@wdc.com> <4993A297653ECB4581FA5C3C31323D195B8D825D@avsrvexchmbx2.microsemi.net> <1503092817.2622.19.camel@wdc.com> <4993A297653ECB4581FA5C3C31323D195B8DBDE4@avsrvexchmbx2.microsemi.net> <1503345191.2571.13.camel@wdc.com> <4993A297653ECB4581FA5C3C31323D195B8DBE34@avsrvexchmbx2.microsemi.net> <4993A297653ECB4581FA5C3C31323D195B8E9103@avsrvexchmbx1.microsemi.net> <1504030401.2653.43.camel@wdc.com> <4993A297653ECB4581FA5C3C31323D195B8E9143@avsrvexchmbx1.microsemi.net> <4993A297653ECB4581FA5C3C31323D195B8E9191@avsrvexchmbx1.microsemi.net> <1504045532.2653.68.camel@wdc.com> <4993A297653ECB4581FA5C3C31323D195B8E9288@avsrvexchmbx1.microsemi.net> <1504047619.2653.75.camel@wdc.com> Mime-Version: 1.0 Content-Type: text/plain Return-path: Received: from aserp1040.oracle.com ([141.146.126.69]:42735 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751240AbdH3BYz (ORCPT ); Tue, 29 Aug 2017 21:24:55 -0400 In-Reply-To: <1504047619.2653.75.camel@wdc.com> (Bart Van Assche's message of "Tue, 29 Aug 2017 23:00:20 +0000") Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Bart Van Assche Cc: "hch@infradead.org" , "viswas.g@microsemi.com" , "gerry.morong@microsemi.com" , "mahesh.rajashekhara@microsemi.com" , "POSWALD@suse.com" , "scott.benesh@microsemi.com" , "don.brace@microsemi.com" , "bader.alisaleh@microsemi.com" , "kevin.barnett@microsemi.com" , "joseph.szczypek@hpe.com" , "scott.teel@microsemi.com" , "jejb@linux.vnet.ibm.com" , "justin.lindley@microsemi.com" , "John.Hall@microsemi.com" Bart, > How about asking these users to create a udev rule instead of directly > modifying max_sectors_kb in sysfs? I looked at this for a bit last week to see if I could come up with an elegant way to accommodate values overridden in sysfs and at the same time honor hardware limits changing. However, it quickly gets messy since some parameters are under the request_queue and some are scsi_disk specific. So that involves having override flags several places. Plus there also the whole re-stacking debacle. So I think I prefer udev for this. -- Martin K. Petersen Oracle Linux Engineering