On Tue, 8 Apr 2025 at 10:09, Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx> wrote: > > > > @@ -2187,11 +2198,12 @@ static int mmc_shutdown(struct mmc_host *host) > > int err = 0; > > > > /* > > - * In a specific case for poweroff notify, we need to resume the card > > - * before we can shutdown it properly. > > + * If the card remains suspended at this point and it was done by using > > + * the sleep-cmd (CMD5), we may need to re-initialize it first, to allow > > + * us to send the preferred poweroff-notification cmd at shutdown. > > */ > > if (mmc_can_poweroff_notify(host->card) && > > - !(host->caps2 & MMC_CAP2_FULL_PWR_CYCLE)) > > + !mmc_host_can_poweroff_notify(host, true)) > > Ooookay, I think I got this logic now. I think it makes sense to make it > more explicit in the comment, though: > > "This is then the case when the card is able to handle poweroff > notifications in general but the host could not initiate those for > suspend." > > Something like this? Well, in my opinion I think this would become a bit too much comments in the code. The rather long function-names "mmc_can_poweroff_notify" (that will change to mmc_card_can_poweroff_notify with your series) and "mmc_host_can_poweroff_notify" are rather self-explanatory, don't you think? Kind regards Uffe
