search for: blk_rq_map_sg

...odge podge of things config: i386-allmodconfig (attached as .config) All related error/warning messages are: block/blk-core.c: In function ?attempt_plug_merge?: block/blk-core.c:1253:9: error: ?struct request? has no member named ?q? drivers/block/cpqarray.c:938:2: warning: passing argument 1 of ?blk_rq_map_sg? from incompatible pointer type [enabled by default] include/linux/blkdev.h:845:12: note: expected ?struct request *? but argument is of type ?struct request_queue *? drivers/block/cpqarray.c:938:2: warning: passing argument 2 of ?blk_rq_map_sg? from incompatible pointer type [enabled by default] i...

...odge podge of things config: i386-allmodconfig (attached as .config) All related error/warning messages are: block/blk-core.c: In function ?attempt_plug_merge?: block/blk-core.c:1253:9: error: ?struct request? has no member named ?q? drivers/block/cpqarray.c:938:2: warning: passing argument 1 of ?blk_rq_map_sg? from incompatible pointer type [enabled by default] include/linux/blkdev.h:845:12: note: expected ?struct request *? but argument is of type ?struct request_queue *? drivers/block/cpqarray.c:938:2: warning: passing argument 2 of ?blk_rq_map_sg? from incompatible pointer type [enabled by default] i...

...blk_mq_start_request(req); > > > > + if (type == VIRTIO_BLK_T_DISCARD || type == > VIRTIO_BLK_T_WRITE_ZEROES) { > > + err = virtblk_setup_discard_write_zeroes(req, unmap); > > + if (err) > > + return BLK_STS_RESOURCE; > > + } > > + > > num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > if (num) { > > if (rq_data_dir(req) == WRITE) > > ...since we still do blk_rq_map_sg() here and num should be != 0. No, while here, we should keep the original logic for READ/WRITE commands.

...blk_mq_start_request(req); > > > > + if (type == VIRTIO_BLK_T_DISCARD || type == > VIRTIO_BLK_T_WRITE_ZEROES) { > > + err = virtblk_setup_discard_write_zeroes(req, unmap); > > + if (err) > > + return BLK_STS_RESOURCE; > > + } > > + > > num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > if (num) { > > if (rq_data_dir(req) == WRITE) > > ...since we still do blk_rq_map_sg() here and num should be != 0. No, while here, we should keep the original logic for READ/WRITE commands.

...LK_T_DISCARD || type == > > > VIRTIO_BLK_T_WRITE_ZEROES) { > > > > + err = virtblk_setup_discard_write_zeroes(req, unmap); > > > > + if (err) > > > > + return BLK_STS_RESOURCE; > > > > + } > > > > + > > > > num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > > > if (num) { > > > > if (rq_data_dir(req) == WRITE) > > > > > > ...since we still do blk_rq_map_sg() here and num should be != 0. > > No, while here, we should keep the original logic for READ/WRITE command...

This patchset implements bio-based IO path for virito-blk to improve performance. Fio test shows it gives, 28%, 24%, 21%, 16% IOPS boost and 32%, 17%, 21%, 16% latency improvement for sequential read/write, random read/write respectively. Asias He (2): block: Add blk_bio_map_sg() helper virtio-blk: Add bio-based IO path for virtio-blk block/blk-merge.c | 63 ++++++++++++++

This patchset implements bio-based IO path for virito-blk to improve performance. Fio test shows it gives, 28%, 24%, 21%, 16% IOPS boost and 32%, 17%, 21%, 16% latency improvement for sequential read/write, random read/write respectively. Asias He (2): block: Add blk_bio_map_sg() helper virtio-blk: Add bio-based IO path for virtio-blk block/blk-merge.c | 63 ++++++++++++++

This allows more requests to fit in the descriptor ring. Copying 1.7M kernel image 100 times (with sync between) Before: totsegs = 55661 totlen = 148859962 avg. 2674 After: totsegs = 36097 totlen = 139439355 avg: 3862 Unfortunately, this coalescing is done at blk_rq_map_sg() which is too late to be optimal: requests have already been limited to the value set by blk_queue_max_hw_segments(). For us, that value reflects the number of sg slots we can handle (ie. after clustering). I suspect other drivers have the same issue. Jens? (This patch is a one liner + tempora...

This allows more requests to fit in the descriptor ring. Copying 1.7M kernel image 100 times (with sync between) Before: totsegs = 55661 totlen = 148859962 avg. 2674 After: totsegs = 36097 totlen = 139439355 avg: 3862 Unfortunately, this coalescing is done at blk_rq_map_sg() which is too late to be optimal: requests have already been limited to the value set by blk_queue_max_hw_segments(). For us, that value reflects the number of sg slots we can handle (ie. after clustering). I suspect other drivers have the same issue. Jens? (This patch is a one liner + tempora...

...t; > > + if (type == VIRTIO_BLK_T_DISCARD || type == > > VIRTIO_BLK_T_WRITE_ZEROES) { > > > + err = virtblk_setup_discard_write_zeroes(req, unmap); > > > + if (err) > > > + return BLK_STS_RESOURCE; > > > + } > > > + > > > num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > > if (num) { > > > if (rq_data_dir(req) == WRITE) > > > > ...since we still do blk_rq_map_sg() here and num should be != 0. > No, while here, we should keep the original logic for READ/WRITE commands. My question is: why...

...how may we get more segments than blk_queue_max_segments? > Is driver expected to validate and drop such requests? > Whee! I'm not alone anymore! I have seen similar issues even on non-mq systems; occasionally I'm hitting this bug in drivers/scsi/scsi_lib.c:scsi_init_io() count = blk_rq_map_sg(req->q, req, sdb->table.sgl); BUG_ON(count > sdb->table.nents); There are actually two problems here: The one is that blk_rq_map_sg() requires a table (ie the last argument), but doesn't have any indications on how large the table is. So one needs to check if the returned number o...

...how may we get more segments than blk_queue_max_segments? > Is driver expected to validate and drop such requests? > Whee! I'm not alone anymore! I have seen similar issues even on non-mq systems; occasionally I'm hitting this bug in drivers/scsi/scsi_lib.c:scsi_init_io() count = blk_rq_map_sg(req->q, req, sdb->table.sgl); BUG_ON(count > sdb->table.nents); There are actually two problems here: The one is that blk_rq_map_sg() requires a table (ie the last argument), but doesn't have any indications on how large the table is. So one needs to check if the returned number o...

...type); vbr->out_hdr.sector = type ? 0 : cpu_to_virtio64(vblk->vdev, blk_rq_pos(req)); - vbr->out_hdr.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); + vbr->out_hdr.u.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); blk_mq_start_request(req); - num = blk_rq_map_sg(hctx->queue, req, vbr->sg); - if (num) { - if (rq_data_dir(req) == WRITE) - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_OUT); - else - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_IN); + if (type == VIRTIO_BLK_T_DISCARD) { + vbr->out_hd...

...type); vbr->out_hdr.sector = type ? 0 : cpu_to_virtio64(vblk->vdev, blk_rq_pos(req)); - vbr->out_hdr.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); + vbr->out_hdr.u.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); blk_mq_start_request(req); - num = blk_rq_map_sg(hctx->queue, req, vbr->sg); - if (num) { - if (rq_data_dir(req) == WRITE) - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_OUT); - else - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_IN); + if (type == VIRTIO_BLK_T_DISCARD) { + vbr->out_hd...

...irtio64(vblk->vdev, blk_rq_pos(req)); > > - vbr->out_hdr.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); > > + vbr->out_hdr.u.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); > > > > blk_mq_start_request(req); > > > > - num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > - if (num) { > > - if (rq_data_dir(req) == WRITE) > > - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, > VIRTIO_BLK_T_OUT); > > - else > > - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, > VIRTIO_BLK_T_...

...irtio64(vblk->vdev, blk_rq_pos(req)); > > - vbr->out_hdr.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); > > + vbr->out_hdr.u.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req)); > > > > blk_mq_start_request(req); > > > > - num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > - if (num) { > > - if (rq_data_dir(req) == WRITE) > > - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, > VIRTIO_BLK_T_OUT); > > - else > > - vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, > VIRTIO_BLK_T_...

...t redhat.com; > jasowang at redhat.com; pbonzini at redhat.com; Wang, Wei W > <wei.w.wang at intel.com> > Subject: Re: [PATCH v4] virtio_blk: add DISCARD and WRIET ZEROES commands > support > > On Tue, May 29, 2018 at 09:42:24AM +0800, Changpeng Liu wrote: > > num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > if (num) { > > - if (rq_data_dir(req) == WRITE) > > + if (rq_data_dir(req) == WRITE || type == VIRTIO_BLK_T_DISCARD > || > > + type == VIRTIO_BLK_T_WRITE_ZEROES) > > vbr->out_hdr.type |= cpu_to_virtio32(vblk->...

...t redhat.com; > jasowang at redhat.com; pbonzini at redhat.com; Wang, Wei W > <wei.w.wang at intel.com> > Subject: Re: [PATCH v4] virtio_blk: add DISCARD and WRIET ZEROES commands > support > > On Tue, May 29, 2018 at 09:42:24AM +0800, Changpeng Liu wrote: > > num = blk_rq_map_sg(hctx->queue, req, vbr->sg); > > if (num) { > > - if (rq_data_dir(req) == WRITE) > > + if (rq_data_dir(req) == WRITE || type == VIRTIO_BLK_T_DISCARD > || > > + type == VIRTIO_BLK_T_WRITE_ZEROES) > > vbr->out_hdr.type |= cpu_to_virtio32(vblk->...

This patchset implements bio-based IO path for virito-blk to improve performance. Fio test shows bio-based IO path gives the following performance improvement: 1) Ramdisk device With bio-based IO path, sequential read/write, random read/write IOPS boost : 28%, 24%, 21%, 16% Latency improvement: 32%, 17%, 21%, 16% 2) Fusion IO device With bio-based IO path, sequential

This patchset implements bio-based IO path for virito-blk to improve performance. Fio test shows bio-based IO path gives the following performance improvement: 1) Ramdisk device With bio-based IO path, sequential read/write, random read/write IOPS boost : 28%, 24%, 21%, 16% Latency improvement: 32%, 17%, 21%, 16% 2) Fusion IO device With bio-based IO path, sequential