| Message ID | 20210328102531.1114535-1-martin.kepplinger@puri.sm (mailing list archive) |
|---|---|
| Headers | show |
| Series | scsi: add runtime PM workaround for SD cardreaders | expand |
On Sun, Mar 28, 2021 at 12:25:27PM +0200, Martin Kepplinger wrote: > hi, > > In short: there are SD cardreaders that send MEDIA_CHANGED on > (runtime) resume. We cannot use runtime PM with these devices as > I/O always fails. I'd like to discuss a way to fix this > or at least allow us to work around this problem: In fact, as far as I know _all_ USB SD card readers send Media Changed notifications on resume. Maybe there are some that don't, but I haven't heard of any. Alan Stern
Am Sonntag, dem 28.03.2021 um 10:58 -0400 schrieb Alan Stern: > On Sun, Mar 28, 2021 at 12:25:27PM +0200, Martin Kepplinger wrote: > > hi, > > > > In short: there are SD cardreaders that send MEDIA_CHANGED on > > (runtime) resume. We cannot use runtime PM with these devices as > > I/O always fails. I'd like to discuss a way to fix this > > or at least allow us to work around this problem: > > In fact, as far as I know _all_ USB SD card readers send Media > Changed > notifications on resume. Maybe there are some that don't, but I > haven't > heard of any. > > Alan Stern that makes me worry less about enabling this for "Generic", "Ultra HS- SD/MMC" then. thanks. it also makes me think about whether sd should implement this even for system-resume (not only runtime resume), but I guess that's a minor issue we could add at any time later. martin
hi, In short: there are SD cardreaders that send MEDIA_CHANGED on (runtime) resume. We cannot use runtime PM with these devices as I/O always fails. I'd like to discuss a way to fix this or at least allow us to work around this problem: For the full background, the discussion started in June 2020 here: https://lore.kernel.org/linux-scsi/20200623111018.31954-1-martin.kepplinger@puri.sm/ I'd appreciate any feedback. Especially: Any naming-preferences for the flags? And is the specific device that I need this workaround for (Generic Ultra HS-SD/MMC, connected via USB) too "generic" maybe? Not sure about what possibilities I'd have here... revision history ---------------- v3: (thank you Bart) * create a new BLIST entry to mark affected devices instead of the sysfs module parameter for sd only. still, only sd implements handling the flag for now. * cc linux-pm list v2: https://lore.kernel.org/linux-scsi/20210112093329.3639-1-martin.kepplinger@puri.sm/ * move module parameter to sd * add Documentation v1: https://lore.kernel.org/linux-scsi/20210111152029.28426-1-martin.kepplinger@puri.sm/T/ Martin Kepplinger (4): scsi: add expecting_media_change flag to error path scsi: devinfo: add new flag BLIST_MEDIA_CHANGE scsi: sd: use expecting_media_change for BLIST_MEDIA_CHANGE devices scsi: devinfo: add BLIST_MEDIA_CHANGE for Ultra HS-SD/MMC usb cardreaders drivers/scsi/scsi_devinfo.c | 1 + drivers/scsi/scsi_error.c | 36 +++++++++++++++++++++++++++++++----- drivers/scsi/sd.c | 23 ++++++++++++++++++++++- include/scsi/scsi_device.h | 1 + include/scsi/scsi_devinfo.h | 6 +++--- 5 files changed, 58 insertions(+), 9 deletions(-)