| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <1389336761.5567.403.camel@haakon3.risingtidesystems.com>
Date: Thu, 09 Jan 2014 22:52:41 -0800
From: "Nicholas A. Bellinger" <nab@...ux-iscsi.org>
To: Sagi Grimberg <sagig@...lanox.com>
Cc: "Nicholas A. Bellinger" <nab@...erainc.com>,
target-devel <target-devel@...r.kernel.org>,
linux-scsi <linux-scsi@...r.kernel.org>,
linux-kernel <linux-kernel@...r.kernel.org>,
"Martin K. Petersen" <martin.petersen@...cle.com>,
Christoph Hellwig <hch@....de>, Hannes Reinecke <hare@...e.de>,
Or Gerlitz <ogerlitz@...lanox.com>
Subject: Re: [PATCH 13/14] target/rd: Add DIF protection into rd_execute_rw
On Thu, 2014-01-09 at 12:32 +0200, Sagi Grimberg wrote:
> On 1/8/2014 10:36 PM, Nicholas A. Bellinger wrote:
> > From: Nicholas Bellinger <nab@...ux-iscsi.org>
> >
> > This patch adds support for DIF protection into rd_execute_rw() code
> > for WRITE/READ I/O using sbc_dif_verify_[write,read]() logic.
> >
> > It also adds rd_get_prot_table() for locating protection SGLs
> > assoicated with the ramdisk backend device.
> >
> > Cc: Martin K. Petersen <martin.petersen@...cle.com>
> > Cc: Christoph Hellwig <hch@....de>
> > Cc: Hannes Reinecke <hare@...e.de>
> > Cc: Sagi Grimberg <sagig@...lanox.com>
> > Cc: Or Gerlitz <ogerlitz@...lanox.com>
> > Signed-off-by: Nicholas Bellinger <nab@...ux-iscsi.org>
> > ---
> > drivers/target/target_core_rd.c | 67 +++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 67 insertions(+)
> >
> > diff --git a/drivers/target/target_core_rd.c b/drivers/target/target_core_rd.c
> > index dd99844..3fd51eb 100644
> > --- a/drivers/target/target_core_rd.c
> > +++ b/drivers/target/target_core_rd.c
> > @@ -363,6 +363,26 @@ static struct rd_dev_sg_table *rd_get_sg_table(struct rd_dev *rd_dev, u32 page)
> > return NULL;
> > }
> >
> > +static struct rd_dev_sg_table *rd_get_prot_table(struct rd_dev *rd_dev, u32 page)
> > +{
> > + struct rd_dev_sg_table *sg_table;
> > + u32 i, sg_per_table = (RD_MAX_ALLOCATION_SIZE /
> > + sizeof(struct scatterlist));
> > +
> > + i = page / sg_per_table;
> > + if (i < rd_dev->sg_prot_count) {
> > + sg_table = &rd_dev->sg_prot_array[i];
> > + if ((sg_table->page_start_offset <= page) &&
> > + (sg_table->page_end_offset >= page))
> > + return sg_table;
> > + }
> > +
> > + pr_err("Unable to locate struct prot rd_dev_sg_table for page: %u\n",
> > + page);
> > +
> > + return NULL;
> > +}
> > +
> > static sense_reason_t
> > rd_execute_rw(struct se_cmd *cmd, struct scatterlist *sgl, u32 sgl_nents,
> > enum dma_data_direction data_direction)
> > @@ -377,6 +397,7 @@ rd_execute_rw(struct se_cmd *cmd, struct scatterlist *sgl, u32 sgl_nents,
> > u32 rd_page;
> > u32 src_len;
> > u64 tmp;
> > + sense_reason_t rc;
> >
> > if (dev->rd_flags & RDF_NULLIO) {
> > target_complete_cmd(cmd, SAM_STAT_GOOD);
> > @@ -399,6 +420,29 @@ rd_execute_rw(struct se_cmd *cmd, struct scatterlist *sgl, u32 sgl_nents,
> > data_direction == DMA_FROM_DEVICE ? "Read" : "Write",
> > cmd->t_task_lba, rd_size, rd_page, rd_offset);
> >
> > + if ((cmd->se_cmd_flags & SCF_PROT) && se_dev->dev_attrib.pi_prot_type &&
> > + data_direction == DMA_TO_DEVICE) {
> > + sector_t sector = cmd->data_length / se_dev->dev_attrib.block_size;
> > + struct rd_dev_sg_table *prot_table;
> > + struct scatterlist *prot_sg;
> > + u32 prot_offset, prot_page;
> > +
> > + tmp = cmd->t_task_lba * se_dev->prot_length;
> > + prot_offset = do_div(tmp, PAGE_SIZE);
> > + prot_page = tmp;
> > +
> > + prot_table = rd_get_prot_table(dev, prot_page);
> > + if (!prot_table)
> > + return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
> > +
> > + prot_sg = &prot_table->sg_table[prot_page - prot_table->page_start_offset];
> > +
> > + rc = sbc_dif_verify_write(cmd, cmd->t_task_lba, sector, 0,
> > + prot_sg, prot_offset);
> > + if (rc)
> > + return rc;
> > + }
> > +
> > src_len = PAGE_SIZE - rd_offset;
> > sg_miter_start(&m, sgl, sgl_nents,
> > data_direction == DMA_FROM_DEVICE ?
> > @@ -460,6 +504,29 @@ rd_execute_rw(struct se_cmd *cmd, struct scatterlist *sgl, u32 sgl_nents,
> > }
> > sg_miter_stop(&m);
> >
> > + if ((cmd->se_cmd_flags & SCF_PROT) && se_dev->dev_attrib.pi_prot_type &&
> > + data_direction == DMA_FROM_DEVICE) {
> > + sector_t sector = cmd->data_length / se_dev->dev_attrib.block_size;
> > + struct rd_dev_sg_table *prot_table;
> > + struct scatterlist *prot_sg;
> > + u32 prot_offset, prot_page;
> > +
> > + tmp = cmd->t_task_lba * se_dev->prot_length;
> > + prot_offset = do_div(tmp, PAGE_SIZE);
> > + prot_page = tmp;
> > +
> > + prot_table = rd_get_prot_table(dev, prot_page);
> > + if (!prot_table)
> > + return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
> > +
> > + prot_sg = &prot_table->sg_table[prot_page - prot_table->page_start_offset];
> > +
> > + rc = sbc_dif_verify_read(cmd, cmd->t_task_lba, sector, 0,
> > + prot_sg, prot_offset);
> > + if (rc)
> > + return rc;
> > + }
> > +
> > target_complete_cmd(cmd, SAM_STAT_GOOD);
> > return 0;
> > }
>
> I wander how we can skip sbc_dif_verify_xxxx if the transport already
> offloaded DIF verify.
> I think that the transport should signal the core layer that it is able
> to offload DIF (ADD/STRIP/PASS/VERIFY), in which case the core should
> turn off the backstore DIF verify emulation to sustain performance.
So IBLOCK + PSCSI backends will need to be a non interleaved protection
PASS for fast path operation, and backend protection emulation is
reserved for RAMDISK and perhaps a special FILEIO full emulation mode.
> I assume that if backstore DIF verify is offloaded as well it does not
> really matter.
That is my assumption of well..
--nab
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists