From: Jae Hyun Yoo Date: Tue, 8 Oct 2019 15:55:25 -0700 Subject: [PATCH 3/5] i2c: aspeed: fix master pending state handling In-Reply-To: References: <20191007231313.4700-1-jae.hyun.yoo@linux.intel.com> <20191007231313.4700-4-jae.hyun.yoo@linux.intel.com> <20191008203157.GA184092@google.com> Message-ID: <8533c682-47ae-2861-6587-19f2501cb0c8@linux.intel.com> List-Id: To: linux-aspeed@lists.ozlabs.org MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit On 10/8/2019 2:54 PM, Brendan Higgins wrote: > On Tue, Oct 8, 2019 at 2:13 PM Jae Hyun Yoo > wrote: >> >> On 10/8/2019 1:31 PM, Brendan Higgins wrote: >>> On Mon, Oct 07, 2019 at 04:13:11PM -0700, Jae Hyun Yoo wrote: >>>> In case of master pending state, it should not trigger the master >>>> command because this H/W is sharing the same data buffer for slave >>>> and master operations, so this commit fixes the issue with making >>>> the master command triggering happen when the state goes to active >>>> state. >>> >>> nit: Makes sense, but can you explain what might happen without your >>> change? >> >> If we don't use this fix, a master command could corrupt data in the >> shared buffer if H/W is still on processing slave operation at the >> timing. > > Right, can you add that to the commit message? Sure, will do that. > Is this trivially reproducible? We might want to submit this > separately as a bugfix. It's very rarely observed. > Actually yeah, can you send this separately as a bugfix? I think we > might want to include this in 5.4. Why not. I can send it separately but this patch series should wait for merging the bug fix to avoid context conflicts. > Wolfram and Joel, what do you think? >