Skip to contentSkip to conversation

seperate max sectors and max hw sectors

PATCH In review Mike Christie 3 participants 7 messages Nov 8, 2005

Subsystems (12)
VFS SCSI USB Block layer +8 more subsystems
lore
Share & tools
lore

Mike Christie

author
#0 lore raw

PATCH 2/10 seperate max sectors and max hw sectors

Seperate max_hw_sectors and max_sectors.

LLDs call blk_queue_max_hw_sectors() to set max_hw_sectors. blk_queue_max_sectors will also set max_sectors to a safe default value.

blk_init_queue still calls blk_queue_max_sectors so if there are any LLDs that do not call blk_queue_max_hw_sectors() and were expecting both the max_sectors and max_hw_sectors to be 255 they do not have to do anything.

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

Signed-off-by Mike Christie <[email protected]>
drivers/block/DAC960.c +1 -1
index 3760edf..43207e5 100644
@@ -2543,7 +2543,7 @@ static boolean DAC960_RegisterBlockDevic
   	RequestQueue->queuedata = Controller;
   	blk_queue_max_hw_segments(RequestQueue, Controller->DriverScatterGatherLimit);
 	blk_queue_max_phys_segments(RequestQueue, Controller->DriverScatterGatherLimit);
-	blk_queue_max_sectors(RequestQueue, Controller->MaxBlocksPerCommand);
+	blk_queue_max_hw_sectors(RequestQueue, Controller->MaxBlocksPerCommand);
 	disk->queue = RequestQueue;
 	sprintf(disk->disk_name, "rd/c%dd%d", Controller->ControllerNumber, n);
 	sprintf(disk->devfs_name, "rd/host%d/target%d", Controller->ControllerNumber, n);
drivers/block/cciss.c +2 -2
index 4827860..32f452c 100644
@@ -1242,7 +1242,7 @@ static void cciss_update_drive_info(int 
 		/* This is a limit in the driver and could be eliminated. */
 		blk_queue_max_phys_segments(disk->queue, MAXSGENTRIES);
 
-		blk_queue_max_sectors(disk->queue, 512);
+		blk_queue_max_hw_sectors(disk->queue, 512);
 
 		disk->queue->queuedata = hba[ctlr];
 

@@ -3141,7 +3141,7 @@ static int __devinit cciss_init_one(stru
 	/* This is a limit in the driver and could be eliminated. */
 	blk_queue_max_phys_segments(q, MAXSGENTRIES);
 
-	blk_queue_max_sectors(q, 512);
+	blk_queue_max_hw_sectors(q, 512);
 
 		q->queuedata = hba[i];
 		sprintf(disk->disk_name, "cciss/c%dd%d", i, j);
drivers/block/floppy.c +1 -1
index 5eadbb9..f9982bb 100644
@@ -4260,7 +4260,7 @@ static int __init floppy_init(void)
 		err = -ENOMEM;
 		goto out_unreg_blkdev;
 	}
-	blk_queue_max_sectors(floppy_queue, 64);
+	blk_queue_max_hw_sectors(floppy_queue, 64);
 
 	blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE,
 			    floppy_find, NULL, NULL);
drivers/block/ll_rw_blk.c +31 -5
index f9b83d2..6fd9121 100644
@@ -241,7 +241,7 @@ void blk_queue_make_request(request_queu
 	q->backing_dev_info.ra_pages = (VM_MAX_READAHEAD * 1024) / PAGE_CACHE_SIZE;
 	q->backing_dev_info.state = 0;
 	q->backing_dev_info.capabilities = BDI_CAP_MAP_COPY;
-	blk_queue_max_sectors(q, MAX_SECTORS);
+	blk_queue_max_sectors(q, SAFE_MAX_SECTORS);
 	blk_queue_hardsect_size(q, 512);
 	blk_queue_dma_alignment(q, 511);
 	blk_queue_congestion_threshold(q);

@@ -563,6 +563,32 @@ void blk_queue_max_sectors(request_queue
 EXPORT_SYMBOL(blk_queue_max_sectors);
 
 /**
+ * blk_queue_max_hw_sectors - set max hw sectors for a request for this queue
+ * @q:  the request queue for the device
+ * @max_sectors:  max sectors in the usual 512b unit
+ *
+ * Description:
+ *    Enables a low level driver to set an upper limit on the size of
+ *    received requests.
+ **/
+void blk_queue_max_hw_sectors(request_queue_t *q, unsigned short max_hw_sectors)
+{
+	if ((max_hw_sectors << 9) < PAGE_CACHE_SIZE) {
+		max_hw_sectors = 1 << (PAGE_CACHE_SHIFT - 9);
+		printk("%s: set to minimum %d\n", __FUNCTION__, max_hw_sectors);
+	}
+
+	if (BLK_DEF_MAX_SECTORS > max_hw_sectors)
+		q->max_hw_sectors = q->max_sectors = max_hw_sectors;
+ 	else {
+		q->max_sectors = BLK_DEF_MAX_SECTORS;
+		q->max_hw_sectors = max_hw_sectors;
+	}
+}
+
+EXPORT_SYMBOL(blk_queue_max_hw_sectors);
+
+/**
  * blk_queue_max_phys_segments - set max phys segments for a request for this queue
  * @q:  the request queue for the device
  * @max_segments:  max number of segments

@@ -659,8 +685,8 @@ EXPORT_SYMBOL(blk_queue_hardsect_size);
 void blk_queue_stack_limits(request_queue_t *t, request_queue_t *b)
 {
 	/* zero is "infinity" */
-	t->max_sectors = t->max_hw_sectors =
-		min_not_zero(t->max_sectors,b->max_sectors);
+	t->max_sectors = min_not_zero(t->max_sectors,b->max_sectors);
+	t->max_hw_sectors = min_not_zero(t->max_hw_sectors,b->max_hw_sectors);
 
 	t->max_phys_segments = min(t->max_phys_segments,b->max_phys_segments);
 	t->max_hw_segments = min(t->max_hw_segments,b->max_hw_segments);

@@ -2147,7 +2173,7 @@ int blk_rq_map_user(request_queue_t *q, 
 	struct bio *bio;
 	int reading;
 
-	if (len > (q->max_sectors << 9))
+	if (len > (q->max_hw_sectors << 9))
 		return -EINVAL;
 	if (!len || !ubuf)
 		return -EINVAL;

@@ -2262,7 +2288,7 @@ int blk_rq_map_kern(request_queue_t *q, 
 {
 	struct bio *bio;
 
-	if (len > (q->max_sectors << 9))
+	if (len > (q->max_hw_sectors << 9))
 		return -EINVAL;
 	if (!len || !kbuf)
 		return -EINVAL;
drivers/block/paride/pd.c +1 -1
index fa49d62..8144d3a 100644
@@ -908,7 +908,7 @@ static int __init pd_init(void)
 	if (!pd_queue)
 		goto out1;
 
-	blk_queue_max_sectors(pd_queue, cluster);
+	blk_queue_max_hw_sectors(pd_queue, cluster);
 
 	if (register_blkdev(major, name))
 		goto out2;
drivers/block/pktcdvd.c +2 -2
index a280e67..527d0a5 100644
@@ -1976,7 +1976,7 @@ static int pkt_open_dev(struct pktcdvd_d
 		 * even if the size is a multiple of the packet size.
 		 */
 		spin_lock_irq(q->queue_lock);
-		blk_queue_max_sectors(q, pd->settings.size);
+		blk_queue_max_hw_sectors(q, pd->settings.size);
 		spin_unlock_irq(q->queue_lock);
 		set_bit(PACKET_WRITABLE, &pd->flags);
 	} else {

@@ -2264,7 +2264,7 @@ static void pkt_init_queue(struct pktcdv
 
 	blk_queue_make_request(q, pkt_make_request);
 	blk_queue_hardsect_size(q, CD_FRAMESIZE);
-	blk_queue_max_sectors(q, PACKET_MAX_SECTORS);
+	blk_queue_max_hw_sectors(q, PACKET_MAX_SECTORS);
 	blk_queue_merge_bvec(q, pkt_merge_bvec);
 	q->queuedata = pd;
 }
drivers/block/ps2esdi.c +1 -1
index 29d1518..905af4d 100644
@@ -411,7 +411,7 @@ static int __init ps2esdi_geninit(void)
 		error = -EBUSY;
 		goto err_out3;
 	}
-	blk_queue_max_sectors(ps2esdi_queue, 128);
+	blk_queue_max_hw_sectors(ps2esdi_queue, 128);
 
 	error = -ENOMEM;
 	for (i = 0; i < ps2esdi_drives; i++) {
drivers/block/scsi_ioctl.c +1 -1
index 382dea7..4e390df 100644
@@ -233,7 +233,7 @@ static int sg_io(struct file *file, requ
 	if (verify_command(file, cmd))
 		return -EPERM;
 
-	if (hdr->dxfer_len > (q->max_sectors << 9))
+	if (hdr->dxfer_len > (q->max_hw_sectors << 9))
 		return -EIO;
 
 	if (hdr->dxfer_len)
drivers/block/ub.c +1 -1
index bfb23d5..d3a55f6 100644
@@ -2331,7 +2331,7 @@ static int ub_probe_lun(struct ub_dev *s
 	blk_queue_max_hw_segments(q, UB_MAX_REQ_SG);
 	blk_queue_max_phys_segments(q, UB_MAX_REQ_SG);
 	blk_queue_segment_boundary(q, 0xffffffff);	/* Dubious. */
-	blk_queue_max_sectors(q, UB_MAX_SECTORS);
+	blk_queue_max_hw_sectors(q, UB_MAX_SECTORS);
 	blk_queue_hardsect_size(q, lun->capacity.bsize);
 
 	q->queuedata = lun;
drivers/block/viodasd.c +1 -1
index 2d518aa..1f88678 100644
@@ -541,7 +541,7 @@ retry:
 	d->disk = g;
 	blk_queue_max_hw_segments(q, VIOMAXBLOCKDMA);
 	blk_queue_max_phys_segments(q, VIOMAXBLOCKDMA);
-	blk_queue_max_sectors(q, VIODASD_MAXSECTORS);
+	blk_queue_max_hw_sectors(q, VIODASD_MAXSECTORS);
 	g->major = VIODASD_MAJOR;
 	g->first_minor = dev_no << PARTITION_SHIFT;
 	if (dev_no >= 26)
drivers/block/xd.c +1 -1
index 68b6d7b..488f1af 100644
@@ -234,7 +234,7 @@ static int __init xd_init(void)
 	}
 
 	/* xd_maxsectors depends on controller - so set after detection */
-	blk_queue_max_sectors(xd_queue, xd_maxsectors);
+	blk_queue_max_hw_sectors(xd_queue, xd_maxsectors);
 
 	for (i = 0; i < xd_drives; i++)
 		add_disk(xd_gendisk[i]);
drivers/cdrom/viocd.c +1 -1
index b519178..10df0bf 100644
@@ -694,7 +694,7 @@ static int viocd_probe(struct vio_dev *v
 			VIOCD_DEVICE_DEVFS "%d", deviceno);
 	blk_queue_max_hw_segments(q, 1);
 	blk_queue_max_phys_segments(q, 1);
-	blk_queue_max_sectors(q, 4096 / 512);
+	blk_queue_max_hw_sectors(q, 4096 / 512);
 	gendisk->queue = q;
 	gendisk->fops = &viocd_fops;
 	gendisk->flags = GENHD_FL_CD|GENHD_FL_REMOVABLE;
drivers/ide/ide-disk.c +1 -1
index 234f5de..acd109e 100644
@@ -917,7 +917,7 @@ static void idedisk_setup (ide_drive_t *
 		if (max_s > hwif->rqsize)
 			max_s = hwif->rqsize;
 
-		blk_queue_max_sectors(drive->queue, max_s);
+		blk_queue_max_hw_sectors(drive->queue, max_s);
 	}
 
 	printk(KERN_INFO "%s: max request size: %dKiB\n", drive->name, drive->queue->max_sectors / 2);
drivers/ide/ide-floppy.c +2 -2
index 29c22fc..f031ee8 100644
@@ -1853,7 +1853,7 @@ static void idefloppy_setup (ide_drive_t
 		set_bit(IDEFLOPPY_ZIP_DRIVE, &floppy->flags);
 		/* This value will be visible in the /proc/ide/hdx/settings */
 		floppy->ticks = IDEFLOPPY_TICKS_DELAY;
-		blk_queue_max_sectors(drive->queue, 64);
+		blk_queue_max_shw_ectors(drive->queue, 64);
 	}
 
 	/*

@@ -1862,7 +1862,7 @@ static void idefloppy_setup (ide_drive_t
 	*      it, so please don't remove this.
 	*/
 	if (strncmp(drive->id->model, "IOMEGA Clik!", 11) == 0) {
-		blk_queue_max_sectors(drive->queue, 64);
+		blk_queue_max_hw_sectors(drive->queue, 64);
 		set_bit(IDEFLOPPY_CLIK_DRIVE, &floppy->flags);
 	}
drivers/ide/ide-probe.c +1 -1
index c1128ae..2578f03 100644
@@ -993,7 +993,7 @@ static int ide_init_queue(ide_drive_t *d
 	}
 	if (hwif->rqsize < max_sectors)
 		max_sectors = hwif->rqsize;
-	blk_queue_max_sectors(q, max_sectors);
+	blk_queue_max_hw_sectors(q, max_sectors);
 
 #ifdef CONFIG_PCI
 	/* When we have an IOMMU, we may have a problem where pci_map_sg()
drivers/ide/legacy/hd.c +1 -1
index 242029c..d989cd5 100644
@@ -721,7 +721,7 @@ static int __init hd_init(void)
 		return -ENOMEM;
 	}
 
-	blk_queue_max_sectors(hd_queue, 255);
+	blk_queue_max_hw_sectors(hd_queue, 255);
 	init_timer(&device_timer);
 	device_timer.function = hd_times_out;
 	blk_queue_hardsect_size(hd_queue, 512);
drivers/md/dm-table.c +1 -1
index a6d3baa..a6f2dc6 100644
@@ -638,7 +638,7 @@ int dm_split_args(int *argc, char ***arg
 static void check_for_valid_limits(struct io_restrictions *rs)
 {
 	if (!rs->max_sectors)
-		rs->max_sectors = MAX_SECTORS;
+		rs->max_sectors = SAFE_MAX_SECTORS;
 	if (!rs->max_phys_segments)
 		rs->max_phys_segments = MAX_PHYS_SEGMENTS;
 	if (!rs->max_hw_segments)
drivers/md/linear.c +1 -1
index 946efef..fa90058 100644
@@ -150,7 +150,7 @@ static int linear_run (mddev_t *mddev)
 		 */
 		if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
 		    mddev->queue->max_sectors > (PAGE_SIZE>>9))
-			blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);
+			blk_queue_max_hw_sectors(mddev->queue, PAGE_SIZE>>9);
 
 		disk->size = rdev->size;
 		mddev->array_size += rdev->size;
drivers/md/multipath.c +2 -2
index c06f447..205ce26 100644
@@ -326,7 +326,7 @@ static int multipath_add_disk(mddev_t *m
 		 */
 			if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
 			    mddev->queue->max_sectors > (PAGE_SIZE>>9))
-				blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);
+				blk_queue_max_hw_sectors(mddev->queue, PAGE_SIZE>>9);
 
 			conf->working_disks++;
 			rdev->raid_disk = path;

@@ -480,7 +480,7 @@ static int multipath_run (mddev_t *mddev
 		 * a merge_bvec_fn to be involved in multipath */
 		if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
 		    mddev->queue->max_sectors > (PAGE_SIZE>>9))
-			blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);
+			blk_queue_max_hw_sectors(mddev->queue, PAGE_SIZE>>9);
 
 		if (!rdev->faulty) 
 			conf->working_disks++;
drivers/md/raid0.c +2 -2
index fece327..e895c3d 100644
@@ -158,7 +158,7 @@ static int create_strip_zones (mddev_t *
 
 		if (rdev1->bdev->bd_disk->queue->merge_bvec_fn &&
 		    mddev->queue->max_sectors > (PAGE_SIZE>>9))
-			blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);
+			blk_queue_max_hw_sectors(mddev->queue, PAGE_SIZE>>9);
 
 		if (!smallest || (rdev1->size <smallest->size))
 			smallest = rdev1;

@@ -284,7 +284,7 @@ static int raid0_run (mddev_t *mddev)
 	       mdname(mddev),
 	       mddev->chunk_size >> 9,
 	       (mddev->chunk_size>>1)-1);
-	blk_queue_max_sectors(mddev->queue, mddev->chunk_size >> 9);
+	blk_queue_max_hw_sectors(mddev->queue, mddev->chunk_size >> 9);
 	blk_queue_segment_boundary(mddev->queue, (mddev->chunk_size>>1) - 1);
 
 	conf = kmalloc(sizeof (raid0_conf_t), GFP_KERNEL);
drivers/md/raid1.c +2 -2
index e16f473..c210022 100644
@@ -947,7 +947,7 @@ static int raid1_add_disk(mddev_t *mddev
 			 */
 			if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
 			    mddev->queue->max_sectors > (PAGE_SIZE>>9))
-				blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);
+				blk_queue_max_hw_sectors(mddev->queue, PAGE_SIZE>>9);
 
 			p->head_position = 0;
 			rdev->raid_disk = mirror;

@@ -1475,7 +1475,7 @@ static int run(mddev_t *mddev)
 		 */
 		if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
 		    mddev->queue->max_sectors > (PAGE_SIZE>>9))
-			blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);
+			blk_queue_max_hw_sectors(mddev->queue, PAGE_SIZE>>9);
 
 		disk->head_position = 0;
 		if (!rdev->faulty && rdev->in_sync)
drivers/message/i2o/i2o_block.c +3 -3
index f283b5b..30a5e92 100644
@@ -1097,11 +1097,11 @@ static int i2o_block_probe(struct device
 	queue->queuedata = i2o_blk_dev;
 
 	blk_queue_max_phys_segments(queue, I2O_MAX_PHYS_SEGMENTS);
-	blk_queue_max_sectors(queue, max_sectors);
+	blk_queue_max_hw_sectors(queue, max_sectors);
 	blk_queue_max_hw_segments(queue, i2o_sg_tablesize(c, body_size));
 
-	osm_debug("max sectors = %d\n", queue->max_phys_segments);
-	osm_debug("phys segments = %d\n", queue->max_sectors);
+	osm_debug("phys segments = %d\n", queue->max_phys_segments);
+	osm_debug("max hw sectors = %d\n", queue->max_hw_sectors);
 	osm_debug("max hw segments = %d\n", queue->max_hw_segments);
 
 	/*
drivers/mmc/mmc_queue.c +1 -1
index 0b9682e..6393859 100644
@@ -139,7 +139,7 @@ int mmc_init_queue(struct mmc_queue *mq,
 
 	blk_queue_prep_rq(mq->queue, mmc_prep_request);
 	blk_queue_bounce_limit(mq->queue, limit);
-	blk_queue_max_sectors(mq->queue, host->max_sectors);
+	blk_queue_max_hw_sectors(mq->queue, host->max_sectors);
 	blk_queue_max_phys_segments(mq->queue, host->max_phys_segs);
 	blk_queue_max_hw_segments(mq->queue, host->max_hw_segs);
 	blk_queue_max_segment_size(mq->queue, host->max_seg_size);
drivers/s390/block/dasd.c +1 -1
index 8fc891a..e3e72ab 100644
@@ -1627,7 +1627,7 @@ dasd_setup_queue(struct dasd_device * de
 
 	blk_queue_hardsect_size(device->request_queue, device->bp_block);
 	max = device->discipline->max_blocks << device->s2b_shift;
-	blk_queue_max_sectors(device->request_queue, max);
+	blk_queue_max_hw_sectors(device->request_queue, max);
 	blk_queue_max_phys_segments(device->request_queue, -1L);
 	blk_queue_max_hw_segments(device->request_queue, -1L);
 	blk_queue_max_segment_size(device->request_queue, -1L);
drivers/s390/char/tape_block.c +1 -1
index 1efc9f2..83ae4e4 100644
@@ -231,7 +231,7 @@ tapeblock_setup_device(struct tape_devic
 		goto cleanup_queue;
 
 	blk_queue_hardsect_size(blkdat->request_queue, TAPEBLOCK_HSEC_SIZE);
-	blk_queue_max_sectors(blkdat->request_queue, TAPEBLOCK_MAX_SEC);
+	blk_queue_max_hw_sectors(blkdat->request_queue, TAPEBLOCK_MAX_SEC);
 	blk_queue_max_phys_segments(blkdat->request_queue, -1L);
 	blk_queue_max_hw_segments(blkdat->request_queue, -1L);
 	blk_queue_max_segment_size(blkdat->request_queue, -1L);
drivers/scsi/ipr.c +1 -1
index fa2cb35..c330b9e 100644
@@ -3212,7 +3212,7 @@ static int ipr_slave_configure(struct sc
 		}
 		if (ipr_is_vset_device(res)) {
 			sdev->timeout = IPR_VSET_RW_TIMEOUT;
-			blk_queue_max_sectors(sdev->request_queue, IPR_VSET_MAX_SECTORS);
+			blk_queue_max_hw_sectors(sdev->request_queue, IPR_VSET_MAX_SECTORS);
 		}
 		if (IPR_IS_DASD_DEVICE(res->cfgte.std_inq_data))
 			sdev->allow_restart = 1;
drivers/scsi/libata-scsi.c +1 -1
index eb604b0..0613d3e 100644
@@ -700,7 +700,7 @@ int ata_scsi_slave_config(struct scsi_de
 			 * do not overwrite sdev->host->max_sectors, since
 			 * other drives on this host may not support LBA48
 			 */
-			blk_queue_max_sectors(sdev->request_queue, 2048);
+			blk_queue_max_hw_sectors(sdev->request_queue, 2048);
 		}
 
 		/*
drivers/scsi/scsi_lib.c +1 -1
index 4afef5c..c68888f 100644
@@ -1508,7 +1508,7 @@ struct request_queue *scsi_alloc_queue(s
 
 	blk_queue_max_hw_segments(q, shost->sg_tablesize);
 	blk_queue_max_phys_segments(q, SCSI_MAX_PHYS_SEGMENTS);
-	blk_queue_max_sectors(q, shost->max_sectors);
+	blk_queue_max_hw_sectors(q, shost->max_sectors);
 	blk_queue_bounce_limit(q, scsi_calculate_bounce_limit(shost));
 	blk_queue_segment_boundary(q, shost->dma_boundary);
 	blk_queue_issue_flush_fn(q, scsi_issue_flush_fn);
drivers/usb/storage/scsiglue.c +3 -3
index 4837524..bf4fce5 100644
@@ -119,7 +119,7 @@ static int slave_configure(struct scsi_d
 	 * increase max_sectors. */
 	if (le16_to_cpu(us->pusb_dev->descriptor.idVendor) == USB_VENDOR_ID_GENESYS &&
 			sdev->request_queue->max_sectors > 64)
-		blk_queue_max_sectors(sdev->request_queue, 64);
+		blk_queue_max_hw_sectors(sdev->request_queue, 64);
 
 	/* We can't put these settings in slave_alloc() because that gets
 	 * called before the device type is known.  Consequently these

@@ -398,7 +398,7 @@ static ssize_t show_max_sectors(struct d
 {
 	struct scsi_device *sdev = to_scsi_device(dev);
 
-	return sprintf(buf, "%u\n", sdev->request_queue->max_sectors);
+	return sprintf(buf, "%u\n", sdev->request_queue->max_hw_sectors);
 }
 
 /* Input routine for the sysfs max_sectors file */

@@ -409,7 +409,7 @@ static ssize_t store_max_sectors(struct 
 	unsigned short ms;
 
 	if (sscanf(buf, "%hu", &ms) > 0 && ms <= SCSI_DEFAULT_MAX_SECTORS) {
-		blk_queue_max_sectors(sdev->request_queue, ms);
+		blk_queue_max_hw_sectors(sdev->request_queue, ms);
 		return strlen(buf);
 	}
 	return -EINVAL;
fs/bio.c +10 -7
index 460554b..d7ed8af 100644
@@ -313,7 +313,8 @@ int bio_get_nr_vecs(struct block_device 
 }
 
 static int __bio_add_page(request_queue_t *q, struct bio *bio, struct page
-			  *page, unsigned int len, unsigned int offset)
+			  *page, unsigned int len, unsigned int offset,
+			  unsigned short max_sectors)
 {
 	int retried_segments = 0;
 	struct bio_vec *bvec;

@@ -327,7 +328,7 @@ static int __bio_add_page(request_queue_
 	if (bio->bi_vcnt >= bio->bi_max_vecs)
 		return 0;
 
-	if (((bio->bi_size + len) >> 9) > q->max_sectors)
+	if (((bio->bi_size + len) >> 9) > max_sectors)
 		return 0;
 
 	/*

@@ -401,8 +402,8 @@ static int __bio_add_page(request_queue_
 int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 		 unsigned int offset)
 {
-	return __bio_add_page(bdev_get_queue(bio->bi_bdev), bio, page,
-			      len, offset);
+	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
+	return __bio_add_page(q, bio, page, len, offset, q->max_sectors);
 }
 
 struct bio_map_data {

@@ -514,7 +515,8 @@ struct bio *bio_copy_user(request_queue_
 			break;
 		}
 
-		if (__bio_add_page(q, bio, page, bytes, 0) < bytes) {
+		if (__bio_add_page(q, bio, page, bytes, 0, q->max_hw_sectors) <
+				   bytes) {
 			ret = -EINVAL;
 			break;
 		}

@@ -628,7 +630,8 @@ static struct bio *__bio_map_user_iov(re
 			/*
 			 * sorry...
 			 */
-			if (__bio_add_page(q, bio, pages[j], bytes, offset) < bytes)
+			if (__bio_add_page(q, bio, pages[j], bytes, offset,
+					   q->max_hw_sectors) < bytes)
 				break;
 
 			len -= bytes;

@@ -802,7 +805,7 @@ static struct bio *__bio_map_kern(reques
 			bytes = len;
 
 		if (__bio_add_page(q, bio, virt_to_page(data), bytes,
-				   offset) < bytes)
+				   offset, q->max_hw_sectors) < bytes)
 			break;
 
 		data += bytes;
include/linux/blkdev.h +3 -1
index 46e927b..e0b47fa 100644
@@ -652,6 +652,7 @@ extern void blk_cleanup_queue(request_qu
 extern void blk_queue_make_request(request_queue_t *, make_request_fn *);
 extern void blk_queue_bounce_limit(request_queue_t *, u64);
 extern void blk_queue_max_sectors(request_queue_t *, unsigned short);
+extern void blk_queue_max_hw_sectors(request_queue_t *, unsigned short);
 extern void blk_queue_max_phys_segments(request_queue_t *, unsigned short);
 extern void blk_queue_max_hw_segments(request_queue_t *, unsigned short);
 extern void blk_queue_max_segment_size(request_queue_t *, unsigned int);

@@ -699,7 +700,8 @@ extern int blkdev_issue_flush(struct blo
 
 #define MAX_PHYS_SEGMENTS 128
 #define MAX_HW_SEGMENTS 128
-#define MAX_SECTORS 255
+#define SAFE_MAX_SECTORS 255
+#define BLK_DEF_MAX_SECTORS 1024
 
 #define MAX_SEGMENT_SIZE	65536
#1 lore raw
Quoted 2 lines · Mike Christie

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

A diffstat would have been a nice addition to this kind of patch.

Signature · 3 lines
Stefan Richter
-=====-=-=-= =-== -=---
http://arcgraph.de/sr/
#2 lore raw
In reply to Mike Christie
Quoted 6 lines · Mike Christie

Seperate max_hw_sectors and max_sectors.

LLDs call blk_queue_max_hw_sectors() to set max_hw_sectors. blk_queue_max_sectors will also set max_sectors to a safe default value.

Quoted 7 lines

blk_init_queue still calls blk_queue_max_sectors so if there are any LLDs that do not call blk_queue_max_hw_sectors() and were expecting both the max_sectors and max_hw_sectors to be 255 they do not have to do anything.

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

->max_sectors will become less of a driver property and more of a block/vm propery, so I think the best way to do this is just to have blk_queue_max_sectors() set ->max_hw_sectors directly and lower ->max_sectors appropriately if it is lower. That also comes with the bonus of not having to modify drivers.

Signature · 1 line
Jens Axboe
author
#3 lore raw
Quoted 17 lines · Jens Axboe
Quoted 6 lines · Mike Christie

Seperate max_hw_sectors and max_sectors.

LLDs call blk_queue_max_hw_sectors() to set max_hw_sectors. blk_queue_max_sectors will also set max_sectors to a safe default value.

Quoted 7 lines

blk_init_queue still calls blk_queue_max_sectors so if there are any LLDs that do not call blk_queue_max_hw_sectors() and were expecting both the max_sectors and max_hw_sectors to be 255 they do not have to do anything.

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

Quoted 6 lines

->max_sectors will become less of a driver property and more of a block/vm propery, so I think the best way to do this is just to have blk_queue_max_sectors() set ->max_hw_sectors directly and lower ->max_sectors appropriately if it is lower. That also comes with the bonus of not having to modify drivers.

Ugggh. I did this in reverse to make the naming nicer. So I added a blk_queue_max_hw_sectors() which sets ->max_sectors to some Block layer default and ->max_hw_sectors to the hw limit (for SCSI this is the scsi host template ->max_sectors). Is this ok? It is more clear for driver writers that they are setting max_hw_sectors when calling blk_queue_max_hw_sectors(). I also converted all the blk_queue_max_sectors() to blk_queue_max_hw_sectors().

#4 lore raw
Quoted 25 lines · Mike Christie
Quoted 17 lines · Jens Axboe
Quoted 6 lines · Mike Christie

Seperate max_hw_sectors and max_sectors.

LLDs call blk_queue_max_hw_sectors() to set max_hw_sectors. blk_queue_max_sectors will also set max_sectors to a safe default value.

Quoted 7 lines

blk_init_queue still calls blk_queue_max_sectors so if there are any LLDs that do not call blk_queue_max_hw_sectors() and were expecting both the max_sectors and max_hw_sectors to be 255 they do not have to do anything.

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

Quoted 6 lines

->max_sectors will become less of a driver property and more of a block/vm propery, so I think the best way to do this is just to have blk_queue_max_sectors() set ->max_hw_sectors directly and lower ->max_sectors appropriately if it is lower. That also comes with the bonus of not having to modify drivers.

Quoted 7 lines

Ugggh. I did this in reverse to make the naming nicer. So I added a blk_queue_max_hw_sectors() which sets ->max_sectors to some Block layer default and ->max_hw_sectors to the hw limit (for SCSI this is the scsi host template ->max_sectors). Is this ok? It is more clear for driver writers that they are setting max_hw_sectors when calling blk_queue_max_hw_sectors(). I also converted all the blk_queue_max_sectors() to blk_queue_max_hw_sectors().

Driver writers need not know. They call blk_queue_max_sectors() to set the maximum value of a request they can handle, they could not care less what internal field in the queue is actually used for that. From their perspective, blk_queue_max_sectors() defines their hard limit. We may (and will) adjust the soft limit behind their back - which is ok, as long as we honor the value they defined by calling blk_queue_max_sectors().

Signature · 1 line
Jens Axboe
author
#5 lore raw
Quoted 34 lines · Jens Axboe
Quoted 25 lines · Mike Christie
Quoted 17 lines · Jens Axboe
Quoted 6 lines · Mike Christie

Seperate max_hw_sectors and max_sectors.

LLDs call blk_queue_max_hw_sectors() to set max_hw_sectors. blk_queue_max_sectors will also set max_sectors to a safe default value.

Quoted 7 lines

blk_init_queue still calls blk_queue_max_sectors so if there are any LLDs that do not call blk_queue_max_hw_sectors() and were expecting both the max_sectors and max_hw_sectors to be 255 they do not have to do anything.

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

Quoted 6 lines

->max_sectors will become less of a driver property and more of a block/vm propery, so I think the best way to do this is just to have blk_queue_max_sectors() set ->max_hw_sectors directly and lower ->max_sectors appropriately if it is lower. That also comes with the bonus of not having to modify drivers.

Quoted 7 lines

Ugggh. I did this in reverse to make the naming nicer. So I added a blk_queue_max_hw_sectors() which sets ->max_sectors to some Block layer default and ->max_hw_sectors to the hw limit (for SCSI this is the scsi host template ->max_sectors). Is this ok? It is more clear for driver writers that they are setting max_hw_sectors when calling blk_queue_max_hw_sectors(). I also converted all the blk_queue_max_sectors() to blk_queue_max_hw_sectors().

Quoted 8 lines

Driver writers need not know. They call blk_queue_max_sectors() to set the maximum value of a request they can handle, they could not care less what internal field in the queue is actually used for that. From their perspective, blk_queue_max_sectors() defines their hard limit. We may (and will) adjust the soft limit behind their back - which is ok, as long as we honor the value they defined by calling blk_queue_max_sectors().

all right, is this patch ok? I changed MAX_SECTORS to SAFE_MAX_SECTORS incase someone was relying on it being 255, and it adds the block layer default to be BLK_DEF_MAX_SECTORS (1024). It also makes block layer SG_IO use max_hw_sectors instead of max_sectors. But what is sg_set_reserved_size used for?

This was made against the first patch in the patchset, but can apply with offsets without it.

drivers/block/ll_rw_blk.c +11 -6
index f9b83d2..b1c4604 100644
@@ -241,7 +241,7 @@ void blk_queue_make_request(request_queu
 	q->backing_dev_info.ra_pages = (VM_MAX_READAHEAD * 1024) / PAGE_CACHE_SIZE;
 	q->backing_dev_info.state = 0;
 	q->backing_dev_info.capabilities = BDI_CAP_MAP_COPY;
-	blk_queue_max_sectors(q, MAX_SECTORS);
+	blk_queue_max_sectors(q, SAFE_MAX_SECTORS);
 	blk_queue_hardsect_size(q, 512);
 	blk_queue_dma_alignment(q, 511);
 	blk_queue_congestion_threshold(q);

@@ -557,7 +557,12 @@ void blk_queue_max_sectors(request_queue
 		printk("%s: set to minimum %d\n", __FUNCTION__, max_sectors);
 	}
 
-	q->max_sectors = q->max_hw_sectors = max_sectors;
+	if (BLK_DEF_MAX_SECTORS > max_hw_sectors)
+		q->max_hw_sectors = q->max_sectors = max_hw_sectors;
+ 	else {
+		q->max_sectors = BLK_DEF_MAX_SECTORS;
+		q->max_hw_sectors = max_hw_sectors;
+	}
 }
 
 EXPORT_SYMBOL(blk_queue_max_sectors);

@@ -659,8 +664,8 @@ EXPORT_SYMBOL(blk_queue_hardsect_size);
 void blk_queue_stack_limits(request_queue_t *t, request_queue_t *b)
 {
 	/* zero is "infinity" */
-	t->max_sectors = t->max_hw_sectors =
-		min_not_zero(t->max_sectors,b->max_sectors);
+	t->max_sectors = min_not_zero(t->max_sectors,b->max_sectors);
+	t->max_hw_sectors = min_not_zero(t->max_hw_sectors,b->max_hw_sectors);
 
 	t->max_phys_segments = min(t->max_phys_segments,b->max_phys_segments);
 	t->max_hw_segments = min(t->max_hw_segments,b->max_hw_segments);

@@ -2147,7 +2152,7 @@ int blk_rq_map_user(request_queue_t *q, 
 	struct bio *bio;
 	int reading;
 
-	if (len > (q->max_sectors << 9))
+	if (len > (q->max_hw_sectors << 9))
 		return -EINVAL;
 	if (!len || !ubuf)
 		return -EINVAL;

@@ -2262,7 +2267,7 @@ int blk_rq_map_kern(request_queue_t *q, 
 {
 	struct bio *bio;
 
-	if (len > (q->max_sectors << 9))
+	if (len > (q->max_hw_sectors << 9))
 		return -EINVAL;
 	if (!len || !kbuf)
 		return -EINVAL;
drivers/block/scsi_ioctl.c +1 -1
index 382dea7..4e390df 100644
@@ -233,7 +233,7 @@ static int sg_io(struct file *file, requ
 	if (verify_command(file, cmd))
 		return -EPERM;
 
-	if (hdr->dxfer_len > (q->max_sectors << 9))
+	if (hdr->dxfer_len > (q->max_hw_sectors << 9))
 		return -EIO;
 
 	if (hdr->dxfer_len)
fs/bio.c +10 -7
index 460554b..d7ed8af 100644
@@ -313,7 +313,8 @@ int bio_get_nr_vecs(struct block_device 
 }
 
 static int __bio_add_page(request_queue_t *q, struct bio *bio, struct page
-			  *page, unsigned int len, unsigned int offset)
+			  *page, unsigned int len, unsigned int offset,
+			  unsigned short max_sectors)
 {
 	int retried_segments = 0;
 	struct bio_vec *bvec;

@@ -327,7 +328,7 @@ static int __bio_add_page(request_queue_
 	if (bio->bi_vcnt >= bio->bi_max_vecs)
 		return 0;
 
-	if (((bio->bi_size + len) >> 9) > q->max_sectors)
+	if (((bio->bi_size + len) >> 9) > max_sectors)
 		return 0;
 
 	/*

@@ -401,8 +402,8 @@ static int __bio_add_page(request_queue_
 int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 		 unsigned int offset)
 {
-	return __bio_add_page(bdev_get_queue(bio->bi_bdev), bio, page,
-			      len, offset);
+	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
+	return __bio_add_page(q, bio, page, len, offset, q->max_sectors);
 }
 
 struct bio_map_data {

@@ -514,7 +515,8 @@ struct bio *bio_copy_user(request_queue_
 			break;
 		}
 
-		if (__bio_add_page(q, bio, page, bytes, 0) < bytes) {
+		if (__bio_add_page(q, bio, page, bytes, 0, q->max_hw_sectors) <
+				   bytes) {
 			ret = -EINVAL;
 			break;
 		}

@@ -628,7 +630,8 @@ static struct bio *__bio_map_user_iov(re
 			/*
 			 * sorry...
 			 */
-			if (__bio_add_page(q, bio, pages[j], bytes, offset) < bytes)
+			if (__bio_add_page(q, bio, pages[j], bytes, offset,
+					   q->max_hw_sectors) < bytes)
 				break;
 
 			len -= bytes;

@@ -802,7 +805,7 @@ static struct bio *__bio_map_kern(reques
 			bytes = len;
 
 		if (__bio_add_page(q, bio, virt_to_page(data), bytes,
-				   offset) < bytes)
+				   offset, q->max_hw_sectors) < bytes)
 			break;
 
 		data += bytes;
include/linux/blkdev.h +2 -1
index 46e927b..37ac77b 100644
@@ -699,7 +699,8 @@ extern int blkdev_issue_flush(struct blo
 
 #define MAX_PHYS_SEGMENTS 128
 #define MAX_HW_SEGMENTS 128
-#define MAX_SECTORS 255
+#define SAFE_MAX_SECTORS 255
+#define BLK_DEF_MAX_SECTORS 1024
 
 #define MAX_SEGMENT_SIZE	65536
author
#6 lore raw
Quoted 44 lines · Mike Christie
Quoted 34 lines · Jens Axboe
Quoted 25 lines · Mike Christie
Quoted 17 lines · Jens Axboe
Quoted 6 lines · Mike Christie

Seperate max_hw_sectors and max_sectors.

LLDs call blk_queue_max_hw_sectors() to set max_hw_sectors. blk_queue_max_sectors will also set max_sectors to a safe default value.

Quoted 7 lines

blk_init_queue still calls blk_queue_max_sectors so if there are any LLDs that do not call blk_queue_max_hw_sectors() and were expecting both the max_sectors and max_hw_sectors to be 255 they do not have to do anything.

I was not able to test every driver I touched, but I think the only place I may have messed up is MD so some testing is needed.

Quoted 6 lines

->max_sectors will become less of a driver property and more of a block/vm propery, so I think the best way to do this is just to have blk_queue_max_sectors() set ->max_hw_sectors directly and lower ->max_sectors appropriately if it is lower. That also comes with the bonus of not having to modify drivers.

Quoted 7 lines

Ugggh. I did this in reverse to make the naming nicer. So I added a blk_queue_max_hw_sectors() which sets ->max_sectors to some Block layer default and ->max_hw_sectors to the hw limit (for SCSI this is the scsi host template ->max_sectors). Is this ok? It is more clear for driver writers that they are setting max_hw_sectors when calling blk_queue_max_hw_sectors(). I also converted all the blk_queue_max_sectors() to blk_queue_max_hw_sectors().

Quoted 8 lines

Driver writers need not know. They call blk_queue_max_sectors() to set the maximum value of a request they can handle, they could not care less what internal field in the queue is actually used for that. From their perspective, blk_queue_max_sectors() defines their hard limit. We may (and will) adjust the soft limit behind their back - which is ok, as long as we honor the value they defined by calling blk_queue_max_sectors().

Quoted 8 lines

all right, is this patch ok? I changed MAX_SECTORS to SAFE_MAX_SECTORS incase someone was relying on it being 255, and it adds the block layer default to be BLK_DEF_MAX_SECTORS (1024). It also makes block layer SG_IO use max_hw_sectors instead of max_sectors. But what is sg_set_reserved_size used for?

This was made against the first patch in the patchset, but can apply with offsets without it.

Messed up combining the two patches. This one actually works.

drivers/block/ll_rw_blk.c +11 -6
index f9b83d2..88caa44 100644
@@ -241,7 +241,7 @@ void blk_queue_make_request(request_queu
 	q->backing_dev_info.ra_pages = (VM_MAX_READAHEAD * 1024) / PAGE_CACHE_SIZE;
 	q->backing_dev_info.state = 0;
 	q->backing_dev_info.capabilities = BDI_CAP_MAP_COPY;
-	blk_queue_max_sectors(q, MAX_SECTORS);
+	blk_queue_max_sectors(q, SAFE_MAX_SECTORS);
 	blk_queue_hardsect_size(q, 512);
 	blk_queue_dma_alignment(q, 511);
 	blk_queue_congestion_threshold(q);

@@ -557,7 +557,12 @@ void blk_queue_max_sectors(request_queue
 		printk("%s: set to minimum %d\n", __FUNCTION__, max_sectors);
 	}
 
-	q->max_sectors = q->max_hw_sectors = max_sectors;
+	if (BLK_DEF_MAX_SECTORS > max_sectors)
+		q->max_hw_sectors = q->max_sectors = max_sectors;
+ 	else {
+		q->max_sectors = BLK_DEF_MAX_SECTORS;
+		q->max_hw_sectors = max_sectors;
+	}
 }
 
 EXPORT_SYMBOL(blk_queue_max_sectors);

@@ -659,8 +664,8 @@ EXPORT_SYMBOL(blk_queue_hardsect_size);
 void blk_queue_stack_limits(request_queue_t *t, request_queue_t *b)
 {
 	/* zero is "infinity" */
-	t->max_sectors = t->max_hw_sectors =
-		min_not_zero(t->max_sectors,b->max_sectors);
+	t->max_sectors = min_not_zero(t->max_sectors,b->max_sectors);
+	t->max_hw_sectors = min_not_zero(t->max_hw_sectors,b->max_hw_sectors);
 
 	t->max_phys_segments = min(t->max_phys_segments,b->max_phys_segments);
 	t->max_hw_segments = min(t->max_hw_segments,b->max_hw_segments);

@@ -2147,7 +2152,7 @@ int blk_rq_map_user(request_queue_t *q, 
 	struct bio *bio;
 	int reading;
 
-	if (len > (q->max_sectors << 9))
+	if (len > (q->max_hw_sectors << 9))
 		return -EINVAL;
 	if (!len || !ubuf)
 		return -EINVAL;

@@ -2262,7 +2267,7 @@ int blk_rq_map_kern(request_queue_t *q, 
 {
 	struct bio *bio;
 
-	if (len > (q->max_sectors << 9))
+	if (len > (q->max_hw_sectors << 9))
 		return -EINVAL;
 	if (!len || !kbuf)
 		return -EINVAL;
drivers/block/scsi_ioctl.c +1 -1
index 382dea7..4e390df 100644
@@ -233,7 +233,7 @@ static int sg_io(struct file *file, requ
 	if (verify_command(file, cmd))
 		return -EPERM;
 
-	if (hdr->dxfer_len > (q->max_sectors << 9))
+	if (hdr->dxfer_len > (q->max_hw_sectors << 9))
 		return -EIO;
 
 	if (hdr->dxfer_len)
drivers/md/dm-table.c +1 -1
index a6d3baa..a6f2dc6 100644
@@ -638,7 +638,7 @@ int dm_split_args(int *argc, char ***arg
 static void check_for_valid_limits(struct io_restrictions *rs)
 {
 	if (!rs->max_sectors)
-		rs->max_sectors = MAX_SECTORS;
+		rs->max_sectors = SAFE_MAX_SECTORS;
 	if (!rs->max_phys_segments)
 		rs->max_phys_segments = MAX_PHYS_SEGMENTS;
 	if (!rs->max_hw_segments)
fs/bio.c +10 -7
index 460554b..d7ed8af 100644
@@ -313,7 +313,8 @@ int bio_get_nr_vecs(struct block_device 
 }
 
 static int __bio_add_page(request_queue_t *q, struct bio *bio, struct page
-			  *page, unsigned int len, unsigned int offset)
+			  *page, unsigned int len, unsigned int offset,
+			  unsigned short max_sectors)
 {
 	int retried_segments = 0;
 	struct bio_vec *bvec;

@@ -327,7 +328,7 @@ static int __bio_add_page(request_queue_
 	if (bio->bi_vcnt >= bio->bi_max_vecs)
 		return 0;
 
-	if (((bio->bi_size + len) >> 9) > q->max_sectors)
+	if (((bio->bi_size + len) >> 9) > max_sectors)
 		return 0;
 
 	/*

@@ -401,8 +402,8 @@ static int __bio_add_page(request_queue_
 int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 		 unsigned int offset)
 {
-	return __bio_add_page(bdev_get_queue(bio->bi_bdev), bio, page,
-			      len, offset);
+	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
+	return __bio_add_page(q, bio, page, len, offset, q->max_sectors);
 }
 
 struct bio_map_data {

@@ -514,7 +515,8 @@ struct bio *bio_copy_user(request_queue_
 			break;
 		}
 
-		if (__bio_add_page(q, bio, page, bytes, 0) < bytes) {
+		if (__bio_add_page(q, bio, page, bytes, 0, q->max_hw_sectors) <
+				   bytes) {
 			ret = -EINVAL;
 			break;
 		}

@@ -628,7 +630,8 @@ static struct bio *__bio_map_user_iov(re
 			/*
 			 * sorry...
 			 */
-			if (__bio_add_page(q, bio, pages[j], bytes, offset) < bytes)
+			if (__bio_add_page(q, bio, pages[j], bytes, offset,
+					   q->max_hw_sectors) < bytes)
 				break;
 
 			len -= bytes;

@@ -802,7 +805,7 @@ static struct bio *__bio_map_kern(reques
 			bytes = len;
 
 		if (__bio_add_page(q, bio, virt_to_page(data), bytes,
-				   offset) < bytes)
+				   offset, q->max_hw_sectors) < bytes)
 			break;
 
 		data += bytes;
include/linux/blkdev.h +2 -1
index 46e927b..37ac77b 100644
@@ -699,7 +699,8 @@ extern int blkdev_issue_flush(struct blo
 
 #define MAX_PHYS_SEGMENTS 128
 #define MAX_HW_SEGMENTS 128
-#define MAX_SECTORS 255
+#define SAFE_MAX_SECTORS 255
+#define BLK_DEF_MAX_SECTORS 1024
 
 #define MAX_SEGMENT_SIZE	65536