From: Martin K. Petersen Date: Thu, 17 Jul 2008 08:28:35 +0000 (-0400) Subject: [SCSI] sd: Support for SCSI disk (SBC) Data Integrity Field X-Git-Tag: v2.6.27-rc1~55^2~19 X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=af55ff675a8461da6a632320710b050af4366e0c;p=linux-2.6 [SCSI] sd: Support for SCSI disk (SBC) Data Integrity Field Support for controllers and disks that implement DIF protection information: - During command preparation the RDPROTECT/WRPROTECT must be set correctly if the target has DIF enabled. - READ(6) and WRITE(6) are not supported when DIF is on. - The controller must be told how to handle the I/O via the protection operation field in scsi_cmnd. - Refactor the I/O completion code that extracts failed LBA from the returned sense data and handle DIF failures correctly. - sd_dif.c implements the functions required to prepare and complete requests with protection information attached. Signed-off-by: Martin K. Petersen Signed-off-by: James Bottomley --- diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig index 26be540d1d..c7f06298bd 100644 --- a/drivers/scsi/Kconfig +++ b/drivers/scsi/Kconfig @@ -63,6 +63,7 @@ comment "SCSI support type (disk, tape, CD-ROM)" config BLK_DEV_SD tristate "SCSI disk support" depends on SCSI + select CRC_T10DIF ---help--- If you want to use SCSI hard disks, Fibre Channel disks, Serial ATA (SATA) or Parallel ATA (PATA) hard disks, diff --git a/drivers/scsi/Makefile b/drivers/scsi/Makefile index a8149677de..72fd5043cf 100644 --- a/drivers/scsi/Makefile +++ b/drivers/scsi/Makefile @@ -151,6 +151,8 @@ scsi_mod-$(CONFIG_SCSI_PROC_FS) += scsi_proc.o scsi_tgt-y += scsi_tgt_lib.o scsi_tgt_if.o sd_mod-objs := sd.o +sd_mod-$(CONFIG_BLK_DEV_INTEGRITY) += sd_dif.o + sr_mod-objs := sr.o sr_ioctl.o sr_vendor.o ncr53c8xx-flags-$(CONFIG_SCSI_ZALON) \ := -DCONFIG_NCR53C8XX_PREFETCH -DSCSI_NCR_BIG_ENDIAN \ diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c index 56b9501d12..8e08d51a0f 100644 --- a/drivers/scsi/sd.c +++ b/drivers/scsi/sd.c @@ -373,6 +373,7 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq) struct scsi_cmnd *SCpnt; struct scsi_device *sdp = q->queuedata; struct gendisk *disk = rq->rq_disk; + struct scsi_disk *sdkp; sector_t block = rq->sector; unsigned int this_count = rq->nr_sectors; unsigned int timeout = sdp->timeout; @@ -389,6 +390,7 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq) if (ret != BLKPREP_OK) goto out; SCpnt = rq->special; + sdkp = scsi_disk(disk); /* from here on until we're complete, any goto out * is used for a killable error condition */ @@ -478,6 +480,11 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq) } SCpnt->cmnd[0] = WRITE_6; SCpnt->sc_data_direction = DMA_TO_DEVICE; + + if (blk_integrity_rq(rq) && + sd_dif_prepare(rq, block, sdp->sector_size) == -EIO) + goto out; + } else if (rq_data_dir(rq) == READ) { SCpnt->cmnd[0] = READ_6; SCpnt->sc_data_direction = DMA_FROM_DEVICE; @@ -492,8 +499,12 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq) "writing" : "reading", this_count, rq->nr_sectors)); - SCpnt->cmnd[1] = 0; - + /* Set RDPROTECT/WRPROTECT if disk is formatted with DIF */ + if (scsi_host_dif_capable(sdp->host, sdkp->protection_type)) + SCpnt->cmnd[1] = 1 << 5; + else + SCpnt->cmnd[1] = 0; + if (block > 0xffffffff) { SCpnt->cmnd[0] += READ_16 - READ_6; SCpnt->cmnd[1] |= blk_fua_rq(rq) ? 0x8 : 0; @@ -511,6 +522,7 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq) SCpnt->cmnd[13] = (unsigned char) this_count & 0xff; SCpnt->cmnd[14] = SCpnt->cmnd[15] = 0; } else if ((this_count > 0xff) || (block > 0x1fffff) || + scsi_device_protection(SCpnt->device) || SCpnt->device->use_10_for_rw) { if (this_count > 0xffff) this_count = 0xffff; @@ -545,6 +557,10 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq) } SCpnt->sdb.length = this_count * sdp->sector_size; + /* If DIF or DIX is enabled, tell HBA how to handle request */ + if (sdkp->protection_type || scsi_prot_sg_count(SCpnt)) + sd_dif_op(SCpnt, sdkp->protection_type, scsi_prot_sg_count(SCpnt)); + /* * We shouldn't disconnect in the middle of a sector, so with a dumb * host adapter, it's safe to assume that we can at least transfer @@ -939,6 +955,48 @@ static struct block_device_operations sd_fops = { .revalidate_disk = sd_revalidate_disk, }; +static unsigned int sd_completed_bytes(struct scsi_cmnd *scmd) +{ + u64 start_lba = scmd->request->sector; + u64 end_lba = scmd->request->sector + (scsi_bufflen(scmd) / 512); + u64 bad_lba; + int info_valid; + + if (!blk_fs_request(scmd->request)) + return 0; + + info_valid = scsi_get_sense_info_fld(scmd->sense_buffer, + SCSI_SENSE_BUFFERSIZE, + &bad_lba); + if (!info_valid) + return 0; + + if (scsi_bufflen(scmd) <= scmd->device->sector_size) + return 0; + + if (scmd->device->sector_size < 512) { + /* only legitimate sector_size here is 256 */ + start_lba <<= 1; + end_lba <<= 1; + } else { + /* be careful ... don't want any overflows */ + u64 factor = scmd->device->sector_size / 512; + do_div(start_lba, factor); + do_div(end_lba, factor); + } + + /* The bad lba was reported incorrectly, we have no idea where + * the error is. + */ + if (bad_lba < start_lba || bad_lba >= end_lba) + return 0; + + /* This computation should always be done in terms of + * the resolution of the device's medium. + */ + return (bad_lba - start_lba) * scmd->device->sector_size; +} + /** * sd_done - bottom half handler: called when the lower level * driver has completed (successfully or otherwise) a scsi command. @@ -949,15 +1007,10 @@ static struct block_device_operations sd_fops = { static int sd_done(struct scsi_cmnd *SCpnt) { int result = SCpnt->result; - unsigned int xfer_size = scsi_bufflen(SCpnt); - unsigned int good_bytes = result ? 0 : xfer_size; - u64 start_lba = SCpnt->request->sector; - u64 end_lba = SCpnt->request->sector + (xfer_size / 512); - u64 bad_lba; + unsigned int good_bytes = result ? 0 : scsi_bufflen(SCpnt); struct scsi_sense_hdr sshdr; int sense_valid = 0; int sense_deferred = 0; - int info_valid; if (result) { sense_valid = scsi_command_normalize_sense(SCpnt, &sshdr); @@ -982,36 +1035,7 @@ static int sd_done(struct scsi_cmnd *SCpnt) switch (sshdr.sense_key) { case HARDWARE_ERROR: case MEDIUM_ERROR: - if (!blk_fs_request(SCpnt->request)) - goto out; - info_valid = scsi_get_sense_info_fld(SCpnt->sense_buffer, - SCSI_SENSE_BUFFERSIZE, - &bad_lba); - if (!info_valid) - goto out; - if (xfer_size <= SCpnt->device->sector_size) - goto out; - if (SCpnt->device->sector_size < 512) { - /* only legitimate sector_size here is 256 */ - start_lba <<= 1; - end_lba <<= 1; - } else { - /* be careful ... don't want any overflows */ - u64 factor = SCpnt->device->sector_size / 512; - do_div(start_lba, factor); - do_div(end_lba, factor); - } - - if (bad_lba < start_lba || bad_lba >= end_lba) - /* the bad lba was reported incorrectly, we have - * no idea where the error is - */ - goto out; - - /* This computation should always be done in terms of - * the resolution of the device's medium. - */ - good_bytes = (bad_lba - start_lba)*SCpnt->device->sector_size; + good_bytes = sd_completed_bytes(SCpnt); break; case RECOVERED_ERROR: case NO_SENSE: @@ -1021,10 +1045,23 @@ static int sd_done(struct scsi_cmnd *SCpnt) scsi_print_sense("sd", SCpnt); SCpnt->result = 0; memset(SCpnt->sense_buffer, 0, SCSI_SENSE_BUFFERSIZE); - good_bytes = xfer_size; + good_bytes = scsi_bufflen(SCpnt); + break; + case ABORTED_COMMAND: + if (sshdr.asc == 0x10) { /* DIF: Disk detected corruption */ + scsi_print_result(SCpnt); + scsi_print_sense("sd", SCpnt); + good_bytes = sd_completed_bytes(SCpnt); + } break; case ILLEGAL_REQUEST: - if (SCpnt->device->use_10_for_rw && + if (sshdr.asc == 0x10) { /* DIX: HBA detected corruption */ + scsi_print_result(SCpnt); + scsi_print_sense("sd", SCpnt); + good_bytes = sd_completed_bytes(SCpnt); + } + if (!scsi_device_protection(SCpnt->device) && + SCpnt->device->use_10_for_rw && (SCpnt->cmnd[0] == READ_10 || SCpnt->cmnd[0] == WRITE_10)) SCpnt->device->use_10_for_rw = 0; @@ -1037,6 +1074,9 @@ static int sd_done(struct scsi_cmnd *SCpnt) break; } out: + if (rq_data_dir(SCpnt->request) == READ && scsi_prot_sg_count(SCpnt)) + sd_dif_complete(SCpnt, good_bytes); + return good_bytes; } @@ -1826,6 +1866,7 @@ static int sd_probe(struct device *dev) dev_set_drvdata(dev, sdkp); add_disk(gd); + sd_dif_config_host(sdkp); sd_printk(KERN_NOTICE, sdkp, "Attached SCSI %sdisk\n", sdp->removable ? "removable " : ""); diff --git a/drivers/scsi/sd.h b/drivers/scsi/sd.h index 86b18d4170..550b2f70a1 100644 --- a/drivers/scsi/sd.h +++ b/drivers/scsi/sd.h @@ -82,4 +82,29 @@ enum sd_dif_target_protection_types { SD_DIF_TYPE3_PROTECTION = 0x3, }; +/* + * Data Integrity Field tuple. + */ +struct sd_dif_tuple { + __be16 guard_tag; /* Checksum */ + __be16 app_tag; /* Opaque storage */ + __be32 ref_tag; /* Target LBA or indirect LBA */ +}; + +#if defined(CONFIG_BLK_DEV_INTEGRITY) + +extern void sd_dif_op(struct scsi_cmnd *, unsigned int, unsigned int); +extern void sd_dif_config_host(struct scsi_disk *); +extern int sd_dif_prepare(struct request *rq, sector_t, unsigned int); +extern void sd_dif_complete(struct scsi_cmnd *, unsigned int); + +#else /* CONFIG_BLK_DEV_INTEGRITY */ + +#define sd_dif_op(a, b, c) do { } while (0) +#define sd_dif_config_host(a) do { } while (0) +#define sd_dif_prepare(a, b, c) (0) +#define sd_dif_complete(a, b) (0) + +#endif /* CONFIG_BLK_DEV_INTEGRITY */ + #endif /* _SCSI_DISK_H */ diff --git a/drivers/scsi/sd_dif.c b/drivers/scsi/sd_dif.c new file mode 100644 index 0000000000..4d17f3d35a --- /dev/null +++ b/drivers/scsi/sd_dif.c @@ -0,0 +1,538 @@ +/* + * sd_dif.c - SCSI Data Integrity Field + * + * Copyright (C) 2007, 2008 Oracle Corporation + * Written by: Martin K. Petersen + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; see the file COPYING. If not, write to + * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, + * USA. + * + */ + +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include "sd.h" + +typedef __u16 (csum_fn) (void *, unsigned int); + +static __u16 sd_dif_crc_fn(void *data, unsigned int len) +{ + return cpu_to_be16(crc_t10dif(data, len)); +} + +static __u16 sd_dif_ip_fn(void *data, unsigned int len) +{ + return ip_compute_csum(data, len); +} + +/* + * Type 1 and Type 2 protection use the same format: 16 bit guard tag, + * 16 bit app tag, 32 bit reference tag. + */ +static void sd_dif_type1_generate(struct blk_integrity_exchg *bix, csum_fn *fn) +{ + void *buf = bix->data_buf; + struct sd_dif_tuple *sdt = bix->prot_buf; + sector_t sector = bix->sector; + unsigned int i; + + for (i = 0 ; i < bix->data_size ; i += bix->sector_size, sdt++) { + sdt->guard_tag = fn(buf, bix->sector_size); + sdt->ref_tag = cpu_to_be32(sector & 0xffffffff); + sdt->app_tag = 0; + + buf += bix->sector_size; + sector++; + } +} + +static void sd_dif_type1_generate_crc(struct blk_integrity_exchg *bix) +{ + sd_dif_type1_generate(bix, sd_dif_crc_fn); +} + +static void sd_dif_type1_generate_ip(struct blk_integrity_exchg *bix) +{ + sd_dif_type1_generate(bix, sd_dif_ip_fn); +} + +static int sd_dif_type1_verify(struct blk_integrity_exchg *bix, csum_fn *fn) +{ + void *buf = bix->data_buf; + struct sd_dif_tuple *sdt = bix->prot_buf; + sector_t sector = bix->sector; + unsigned int i; + __u16 csum; + + for (i = 0 ; i < bix->data_size ; i += bix->sector_size, sdt++) { + /* Unwritten sectors */ + if (sdt->app_tag == 0xffff) + return 0; + + /* Bad ref tag received from disk */ + if (sdt->ref_tag == 0xffffffff) { + printk(KERN_ERR + "%s: bad phys ref tag on sector %lu\n", + bix->disk_name, (unsigned long)sector); + return -EIO; + } + + if (be32_to_cpu(sdt->ref_tag) != (sector & 0xffffffff)) { + printk(KERN_ERR + "%s: ref tag error on sector %lu (rcvd %u)\n", + bix->disk_name, (unsigned long)sector, + be32_to_cpu(sdt->ref_tag)); + return -EIO; + } + + csum = fn(buf, bix->sector_size); + + if (sdt->guard_tag != csum) { + printk(KERN_ERR "%s: guard tag error on sector %lu " \ + "(rcvd %04x, data %04x)\n", bix->disk_name, + (unsigned long)sector, + be16_to_cpu(sdt->guard_tag), be16_to_cpu(csum)); + return -EIO; + } + + buf += bix->sector_size; + sector++; + } + + return 0; +} + +static int sd_dif_type1_verify_crc(struct blk_integrity_exchg *bix) +{ + return sd_dif_type1_verify(bix, sd_dif_crc_fn); +} + +static int sd_dif_type1_verify_ip(struct blk_integrity_exchg *bix) +{ + return sd_dif_type1_verify(bix, sd_dif_ip_fn); +} + +/* + * Functions for interleaving and deinterleaving application tags + */ +static void sd_dif_type1_set_tag(void *prot, void *tag_buf, unsigned int sectors) +{ + struct sd_dif_tuple *sdt = prot; + char *tag = tag_buf; + unsigned int i, j; + + for (i = 0, j = 0 ; i < sectors ; i++, j += 2, sdt++) { + sdt->app_tag = tag[j] << 8 | tag[j+1]; + BUG_ON(sdt->app_tag == 0xffff); + } +} + +static void sd_dif_type1_get_tag(void *prot, void *tag_buf, unsigned int sectors) +{ + struct sd_dif_tuple *sdt = prot; + char *tag = tag_buf; + unsigned int i, j; + + for (i = 0, j = 0 ; i < sectors ; i++, j += 2, sdt++) { + tag[j] = (sdt->app_tag & 0xff00) >> 8; + tag[j+1] = sdt->app_tag & 0xff; + } +} + +static struct blk_integrity dif_type1_integrity_crc = { + .name = "T10-DIF-TYPE1-CRC", + .generate_fn = sd_dif_type1_generate_crc, + .verify_fn = sd_dif_type1_verify_crc, + .get_tag_fn = sd_dif_type1_get_tag, + .set_tag_fn = sd_dif_type1_set_tag, + .tuple_size = sizeof(struct sd_dif_tuple), + .tag_size = 0, +}; + +static struct blk_integrity dif_type1_integrity_ip = { + .name = "T10-DIF-TYPE1-IP", + .generate_fn = sd_dif_type1_generate_ip, + .verify_fn = sd_dif_type1_verify_ip, + .get_tag_fn = sd_dif_type1_get_tag, + .set_tag_fn = sd_dif_type1_set_tag, + .tuple_size = sizeof(struct sd_dif_tuple), + .tag_size = 0, +}; + + +/* + * Type 3 protection has a 16-bit guard tag and 16 + 32 bits of opaque + * tag space. + */ +static void sd_dif_type3_generate(struct blk_integrity_exchg *bix, csum_fn *fn) +{ + void *buf = bix->data_buf; + struct sd_dif_tuple *sdt = bix->prot_buf; + unsigned int i; + + for (i = 0 ; i < bix->data_size ; i += bix->sector_size, sdt++) { + sdt->guard_tag = fn(buf, bix->sector_size); + sdt->ref_tag = 0; + sdt->app_tag = 0; + + buf += bix->sector_size; + } +} + +static void sd_dif_type3_generate_crc(struct blk_integrity_exchg *bix) +{ + sd_dif_type3_generate(bix, sd_dif_crc_fn); +} + +static void sd_dif_type3_generate_ip(struct blk_integrity_exchg *bix) +{ + sd_dif_type3_generate(bix, sd_dif_ip_fn); +} + +static int sd_dif_type3_verify(struct blk_integrity_exchg *bix, csum_fn *fn) +{ + void *buf = bix->data_buf; + struct sd_dif_tuple *sdt = bix->prot_buf; + sector_t sector = bix->sector; + unsigned int i; + __u16 csum; + + for (i = 0 ; i < bix->data_size ; i += bix->sector_size, sdt++) { + /* Unwritten sectors */ + if (sdt->app_tag == 0xffff && sdt->ref_tag == 0xffffffff) + return 0; + + csum = fn(buf, bix->sector_size); + + if (sdt->guard_tag != csum) { + printk(KERN_ERR "%s: guard tag error on sector %lu " \ + "(rcvd %04x, data %04x)\n", bix->disk_name, + (unsigned long)sector, + be16_to_cpu(sdt->guard_tag), be16_to_cpu(csum)); + return -EIO; + } + + buf += bix->sector_size; + sector++; + } + + return 0; +} + +static int sd_dif_type3_verify_crc(struct blk_integrity_exchg *bix) +{ + return sd_dif_type3_verify(bix, sd_dif_crc_fn); +} + +static int sd_dif_type3_verify_ip(struct blk_integrity_exchg *bix) +{ + return sd_dif_type3_verify(bix, sd_dif_ip_fn); +} + +static void sd_dif_type3_set_tag(void *prot, void *tag_buf, unsigned int sectors) +{ + struct sd_dif_tuple *sdt = prot; + char *tag = tag_buf; + unsigned int i, j; + + for (i = 0, j = 0 ; i < sectors ; i++, j += 6, sdt++) { + sdt->app_tag = tag[j] << 8 | tag[j+1]; + sdt->ref_tag = tag[j+2] << 24 | tag[j+3] << 16 | + tag[j+4] << 8 | tag[j+5]; + } +} + +static void sd_dif_type3_get_tag(void *prot, void *tag_buf, unsigned int sectors) +{ + struct sd_dif_tuple *sdt = prot; + char *tag = tag_buf; + unsigned int i, j; + + for (i = 0, j = 0 ; i < sectors ; i++, j += 2, sdt++) { + tag[j] = (sdt->app_tag & 0xff00) >> 8; + tag[j+1] = sdt->app_tag & 0xff; + tag[j+2] = (sdt->ref_tag & 0xff000000) >> 24; + tag[j+3] = (sdt->ref_tag & 0xff0000) >> 16; + tag[j+4] = (sdt->ref_tag & 0xff00) >> 8; + tag[j+5] = sdt->ref_tag & 0xff; + BUG_ON(sdt->app_tag == 0xffff || sdt->ref_tag == 0xffffffff); + } +} + +static struct blk_integrity dif_type3_integrity_crc = { + .name = "T10-DIF-TYPE3-CRC", + .generate_fn = sd_dif_type3_generate_crc, + .verify_fn = sd_dif_type3_verify_crc, + .get_tag_fn = sd_dif_type3_get_tag, + .set_tag_fn = sd_dif_type3_set_tag, + .tuple_size = sizeof(struct sd_dif_tuple), + .tag_size = 0, +}; + +static struct blk_integrity dif_type3_integrity_ip = { + .name = "T10-DIF-TYPE3-IP", + .generate_fn = sd_dif_type3_generate_ip, + .verify_fn = sd_dif_type3_verify_ip, + .get_tag_fn = sd_dif_type3_get_tag, + .set_tag_fn = sd_dif_type3_set_tag, + .tuple_size = sizeof(struct sd_dif_tuple), + .tag_size = 0, +}; + +/* + * Configure exchange of protection information between OS and HBA. + */ +void sd_dif_config_host(struct scsi_disk *sdkp) +{ + struct scsi_device *sdp = sdkp->device; + struct gendisk *disk = sdkp->disk; + u8 type = sdkp->protection_type; + + /* If this HBA doesn't support DIX, resort to normal I/O or DIF */ + if (scsi_host_dix_capable(sdp->host, type) == 0) { + + if (type == SD_DIF_TYPE0_PROTECTION) + return; + + if (scsi_host_dif_capable(sdp->host, type) == 0) { + sd_printk(KERN_INFO, sdkp, "Type %d protection " \ + "unsupported by HBA. Disabling DIF.\n", type); + sdkp->protection_type = 0; + return; + } + + sd_printk(KERN_INFO, sdkp, "Enabling DIF Type %d protection\n", + type); + + return; + } + + /* Enable DMA of protection information */ + if (scsi_host_get_guard(sdkp->device->host) & SHOST_DIX_GUARD_IP) + if (type == SD_DIF_TYPE3_PROTECTION) + blk_integrity_register(disk, &dif_type3_integrity_ip); + else + blk_integrity_register(disk, &dif_type1_integrity_ip); + else + if (type == SD_DIF_TYPE3_PROTECTION) + blk_integrity_register(disk, &dif_type3_integrity_crc); + else + blk_integrity_register(disk, &dif_type1_integrity_crc); + + sd_printk(KERN_INFO, sdkp, + "Enabling %s integrity protection\n", disk->integrity->name); + + /* Signal to block layer that we support sector tagging */ + if (type && sdkp->ATO) { + if (type == SD_DIF_TYPE3_PROTECTION) + disk->integrity->tag_size = sizeof(u16) + sizeof(u32); + else + disk->integrity->tag_size = sizeof(u16); + + sd_printk(KERN_INFO, sdkp, "DIF application tag size %u\n", + disk->integrity->tag_size); + } +} + +/* + * DIF DMA operation magic decoder ring. + */ +void sd_dif_op(struct scsi_cmnd *scmd, unsigned int dif, unsigned int dix) +{ + int csum_convert, prot_op; + + prot_op = 0; + + /* Convert checksum? */ + if (scsi_host_get_guard(scmd->device->host) != SHOST_DIX_GUARD_CRC) + csum_convert = 1; + else + csum_convert = 0; + + switch (scmd->cmnd[0]) { + case READ_10: + case READ_12: + case READ_16: + if (dif && dix) + if (csum_convert) + prot_op = SCSI_PROT_READ_CONVERT; + else + prot_op = SCSI_PROT_READ_PASS; + else if (dif && !dix) + prot_op = SCSI_PROT_READ_STRIP; + else if (!dif && dix) + prot_op = SCSI_PROT_READ_INSERT; + + break; + + case WRITE_10: + case WRITE_12: + case WRITE_16: + if (dif && dix) + if (csum_convert) + prot_op = SCSI_PROT_WRITE_CONVERT; + else + prot_op = SCSI_PROT_WRITE_PASS; + else if (dif && !dix) + prot_op = SCSI_PROT_WRITE_INSERT; + else if (!dif && dix) + prot_op = SCSI_PROT_WRITE_STRIP; + + break; + } + + scsi_set_prot_op(scmd, prot_op); + scsi_set_prot_type(scmd, dif); +} + +/* + * The virtual start sector is the one that was originally submitted + * by the block layer. Due to partitioning, MD/DM cloning, etc. the + * actual physical start sector is likely to be different. Remap + * protection information to match the physical LBA. + * + * From a protocol perspective there's a slight difference between + * Type 1 and 2. The latter uses 32-byte CDBs exclusively, and the + * reference tag is seeded in the CDB. This gives us the potential to + * avoid virt->phys remapping during write. However, at read time we + * don't know whether the virt sector is the same as when we wrote it + * (we could be reading from real disk as opposed to MD/DM device. So + * we always remap Type 2 making it identical to Type 1. + * + * Type 3 does not have a reference tag so no remapping is required. + */ +int sd_dif_prepare(struct request *rq, sector_t hw_sector, unsigned int sector_sz) +{ + const int tuple_sz = sizeof(struct sd_dif_tuple); + struct bio *bio; + struct scsi_disk *sdkp; + struct sd_dif_tuple *sdt; + unsigned int i, j; + u32 phys, virt; + + /* Already remapped? */ + if (rq->cmd_flags & REQ_INTEGRITY) + return 0; + + sdkp = rq->bio->bi_bdev->bd_disk->private_data; + + if (sdkp->protection_type == SD_DIF_TYPE3_PROTECTION) + return 0; + + rq->cmd_flags |= REQ_INTEGRITY; + phys = hw_sector & 0xffffffff; + + __rq_for_each_bio(bio, rq) { + struct bio_vec *iv; + + virt = bio->bi_integrity->bip_sector & 0xffffffff; + + bip_for_each_vec(iv, bio->bi_integrity, i) { + sdt = kmap_atomic(iv->bv_page, KM_USER0) + + iv->bv_offset; + + for (j = 0 ; j < iv->bv_len ; j += tuple_sz, sdt++) { + + if (be32_to_cpu(sdt->ref_tag) != virt) + goto error; + + sdt->ref_tag = cpu_to_be32(phys); + virt++; + phys++; + } + + kunmap_atomic(sdt, KM_USER0); + } + } + + return 0; + +error: + kunmap_atomic(sdt, KM_USER0); + sd_printk(KERN_ERR, sdkp, "%s: virt %u, phys %u, ref %u\n", + __func__, virt, phys, be32_to_cpu(sdt->ref_tag)); + + return -EIO; +} + +/* + * Remap physical sector values in the reference tag to the virtual + * values expected by the block layer. + */ +void sd_dif_complete(struct scsi_cmnd *scmd, unsigned int good_bytes) +{ + const int tuple_sz = sizeof(struct sd_dif_tuple); + struct scsi_disk *sdkp; + struct bio *bio; + struct sd_dif_tuple *sdt; + unsigned int i, j, sectors, sector_sz; + u32 phys, virt; + + sdkp = scsi_disk(scmd->request->rq_disk); + + if (sdkp->protection_type == SD_DIF_TYPE3_PROTECTION || good_bytes == 0) + return; + + sector_sz = scmd->device->sector_size; + sectors = good_bytes / sector_sz; + + phys = scmd->request->sector & 0xffffffff; + if (sector_sz == 4096) + phys >>= 3; + + __rq_for_each_bio(bio, scmd->request) { + struct bio_vec *iv; + + virt = bio->bi_integrity->bip_sector & 0xffffffff; + + bip_for_each_vec(iv, bio->bi_integrity, i) { + sdt = kmap_atomic(iv->bv_page, KM_USER0) + + iv->bv_offset; + + for (j = 0 ; j < iv->bv_len ; j += tuple_sz, sdt++) { + + if (sectors == 0) { + kunmap_atomic(sdt, KM_USER0); + return; + } + + if (be32_to_cpu(sdt->ref_tag) != phys && + sdt->app_tag != 0xffff) + sdt->ref_tag = 0xffffffff; /* Bad ref */ + else + sdt->ref_tag = cpu_to_be32(virt); + + virt++; + phys++; + sectors--; + } + + kunmap_atomic(sdt, KM_USER0); + } + } +} +