On 2021/09/27 18:10, Reimar Döffinger wrote: > >> On 27 Sep 2021, at 10:56, Paul Menzel <pmenzel@xxxxxxxxxxxxx> wrote: >> >> Dear Reimar, >> >> >> Thank you for the patch. >> >> >> Am 19.08.21 um 10:13 schrieb Reimar Döffinger: >> >> Maybe start with a problem statement: >> >> With some SSDs Linux logs the error below: >> >> failed to set xfermode (err_mask=0x40) >> >>> Checking if DMA is enabled should be done via the >>> ata_dma_enabled helper function, since the init state >>> 0xff indicates disabled. >>> Only the libata-core logic is tested on actual devices, >>> the other changes are based on code review only. >> >> Your Signed-off-by line is missing, and you might want to add the Linux kernel bug tracker entry: >> >> BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=195895 > > Thanks, I missed that the Signed-off-by got lost, and thanks for testing. > I actually wanted to ask (e.g. Damien?), would you like me to split out the fully > tested and confirmed working and necessary libata-core change > from the other changes? It is generally good practice to have one patch per driver, preceded if needed by core changes. So yes, please split things if it can be done cleanly. > I am quite confident that all the code modified was wrong before, > however it could be that some of the code actually relies > on that bug to cancel out further bugs, so fixing could well break more > than it fixes... I had a look when you sent the patch and did not see anything scary. Please resend and I will check again. -- Damien Le Moal Western Digital Research