Hi christoph, Thanks for your comments. As mentioned in the cover letter that these ops are needed to implement clock scaling and partial init features for which we already had below discussions but faced strong resistance from community. Since these were huge code changes so maintainability was the main concern. Hence we have redesigned our entire logic and moved complete code to vendor specific file and to support this new design now we just need these two hooks in suspend and resume functions along with few symbols to be exported so that we can use those symbols in our vendor files. I will push the vendor specific changes in the next patchset. Old discussion for Clock scaling feature : https://patchwork.kernel.org/project/linux-mmc/cover/1571668177-3766-1-git-send-email-rampraka@xxxxxxxxxxxxxx/ Old discussion for Partial init feature : https://patchwork.kernel.org/project/linux-mmc/patch/1650963852-4173-1-git-send-email-quic_spathi@xxxxxxxxxxx/ Thanks, Sarthak > -----Original Message----- > From: Christoph Hellwig <hch@xxxxxxxxxxxxx> > Sent: Tuesday, April 4, 2023 10:44 AM > To: Sarthak Garg (QUIC) <quic_sartgarg@xxxxxxxxxxx> > Cc: adrian.hunter@xxxxxxxxx; ulf.hansson@xxxxxxxxxx; linux- > mmc@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; linux-arm- > msm@xxxxxxxxxxxxxxx; Ram Prakash Gupta (QUIC) > <quic_rampraka@xxxxxxxxxxx>; Bhaskar Valaboju (QUIC) > <quic_bhaskarv@xxxxxxxxxxx>; Sachin Gupta (QUIC) > <quic_sachgupt@xxxxxxxxxxx>; Pradeep Pragallapati (QUIC) > <quic_pragalla@xxxxxxxxxxx>; Sayali Lokhande (QUIC) > <quic_sayalil@xxxxxxxxxxx>; Brian Norris <briannorris@xxxxxxxxxxxx>; > Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx>; Linus Walleij > <linus.walleij@xxxxxxxxxx> > Subject: Re: [PATCH V1 1/2] mmc: core: Define new vendor ops to enable > internal features > > On Sat, Apr 01, 2023 at 10:27:22PM +0530, Sarthak Garg wrote: > > Define new ops to let vendor enable internal features in > > mmc_suspend/resume paths like partial init feature. > > 1) vendors have absolutely no business doing anything, you might be > doing either something entirely wrong or use the wrong terminology > here. > > 2) any kind of core hook not only needs a very good description, but > also an actual user that goes along in the same series.
