On Sun, Dec 29, 2019 at 05:15:49PM -0800, Bart Van Assche wrote: > On 2019-12-29 08:14, Jens Axboe wrote: > > On 12/28/19 7:32 PM, Ming Lei wrote: > >> There are two issues in get_max_segment_size(): > >> > >> 1) the default segment boudary mask is bypassed, and some devices still > >> require segment to not cross the default 4G boundary > >> > >> 2) the segment start address isn't taken into account when checking > >> segment boundary limit > >> > >> Fixes the two issues. > > > > Given the potential severity of the bug, I think it deserves a somewhat > > richer explanation than just that. It should also go into stable. This > > is what I queued up: > > > > https://git.kernel.dk/cgit/linux-block/commit/?h=block-5.5&id=add1fc07334260253dfa880d9c964edc8381deac > > Although this patch looks fine to me, seeing Jens' patch description > makes me wonder whether the DMA segment boundary for the mpt3sas adapter > should be made explicit, e.g. by setting the SCSI host .dma_boundary > parameter in the mpt3sas driver? From the SCSI core: > > blk_queue_segment_boundary(q, shost->dma_boundary); See scsi_host_alloc(): ... /* * assume a 4GB boundary, if not set */ if (sht->dma_boundary) shost->dma_boundary = sht->dma_boundary; else shost->dma_boundary = 0xffffffff; Thanks, Ming
