[PATCH][REPOST] lpfc: Fix bad memory access during VPD DUMP mailbox command

[PATCH][REPOST] lpfc: Fix bad memory access during VPD DUMP mailbox command

[James Smart]

The dump command for reading a region passes a requested read length
specified in words (4byte units). The response overwrites the same
field with the actual number of bytes read.

The mailbox handler for DUMP which reads VPD data (region 23) is
treating the response field as if it were still a word_cnt, thus
multiplying it by 4 to set the read's "length". Given the read value
was calculated based on the size of the read buffer, the longer
response length runs off the end of the buffer.

Fix by reworking the code to use the response field as a byte count.

Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>

---
 drivers/scsi/lpfc/lpfc_init.c | 12 ++++++------
 drivers/scsi/lpfc/lpfc_sli.c  | 15 ++++++++-------
 2 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c
index 1e4c792bb660..5f018d02bf56 100644
--- a/drivers/scsi/lpfc/lpfc_init.c
+++ b/drivers/scsi/lpfc/lpfc_init.c
@@ -254,13 +254,13 @@ lpfc_config_port_prep(struct lpfc_hba *phba)
 		if (mb->un.varDmp.word_cnt == 0)
 			break;
 
-		i =  mb->un.varDmp.word_cnt * sizeof(uint32_t);
-		if (offset + i >  DMP_VPD_SIZE)
-			i =  DMP_VPD_SIZE - offset;
+		if (mb->un.varDmp.word_cnt > DMP_VPD_SIZE - offset)
+			mb->un.varDmp.word_cnt = DMP_VPD_SIZE - offset;
 		lpfc_sli_pcimem_bcopy(((uint8_t *)mb) + DMP_RSP_OFFSET,
-				      lpfc_vpd_data  + offset, i);
-		offset += i;
-	} while (offset < DMP_VPD_SIZE);
+				      lpfc_vpd_data + offset,
+				      mb->un.varDmp.word_cnt);
+		offset += mb->un.varDmp.word_cnt;
+	} while (mb->un.varDmp.word_cnt && offset < DMP_VPD_SIZE);
 
 	lpfc_parse_vpd(phba, lpfc_vpd_data, offset);
 
diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/drivers/scsi/lpfc/lpfc_sli.c
index 579ac75dfe79..573c8599d71c 100644
--- a/drivers/scsi/lpfc/lpfc_sli.c
+++ b/drivers/scsi/lpfc/lpfc_sli.c
@@ -19777,7 +19777,7 @@ lpfc_sli_get_config_region23(struct lpfc_hba *phba, char *rgn23_data)
 	LPFC_MBOXQ_t *pmb = NULL;
 	MAILBOX_t *mb;
 	uint32_t offset = 0;
-	int i, rc;
+	int rc;
 
 	if (!rgn23_data)
 		return 0;
@@ -19808,13 +19808,14 @@ lpfc_sli_get_config_region23(struct lpfc_hba *phba, char *rgn23_data)
 		if (mb->un.varDmp.word_cnt == 0)
 			break;
 
-		i =  mb->un.varDmp.word_cnt * sizeof(uint32_t);
-		if (offset + i >  DMP_RGN23_SIZE)
-			i =  DMP_RGN23_SIZE - offset;
+		if (mb->un.varDmp.word_cnt > DMP_RGN23_SIZE - offset)
+			mb->un.varDmp.word_cnt = DMP_RGN23_SIZE - offset;
+
 		lpfc_sli_pcimem_bcopy(((uint8_t *)mb) + DMP_RSP_OFFSET,
-				      rgn23_data  + offset, i);
-		offset += i;
-	} while (offset < DMP_RGN23_SIZE);
+				       rgn23_data + offset,
+				       mb->un.varDmp.word_cnt);
+		offset += mb->un.varDmp.word_cnt;
+	} while (mb->un.varDmp.word_cnt && offset < DMP_RGN23_SIZE);
 
 	mempool_free(pmb, phba->mbox_mem_pool);
 	return offset;
-- 
2.26.2

Re: [PATCH][REPOST] lpfc: Fix bad memory access during VPD DUMP mailbox command

[Martin K. Petersen]

On Wed, 21 Apr 2021 16:45:11 -0700, James Smart wrote:

> The dump command for reading a region passes a requested read length
> specified in words (4byte units). The response overwrites the same
> field with the actual number of bytes read.
> 
> The mailbox handler for DUMP which reads VPD data (region 23) is
> treating the response field as if it were still a word_cnt, thus
> multiplying it by 4 to set the read's "length". Given the read value
> was calculated based on the size of the read buffer, the longer
> response length runs off the end of the buffer.
> 
> [...]

Applied to 5.13/scsi-fixes, thanks!

[1/1] lpfc: Fix bad memory access during VPD DUMP mailbox command
      https://git.kernel.org/mkp/scsi/c/e4ec10228fdf

-- 
Martin K. Petersen	Oracle Linux Engineering

Re: [PATCH][REPOST] lpfc: Fix bad memory access during VPD DUMP mailbox command

[Martin K. Petersen]

On Wed, 21 Apr 2021 16:45:11 -0700, James Smart wrote:

> The dump command for reading a region passes a requested read length
> specified in words (4byte units). The response overwrites the same
> field with the actual number of bytes read.
> 
> The mailbox handler for DUMP which reads VPD data (region 23) is
> treating the response field as if it were still a word_cnt, thus
> multiplying it by 4 to set the read's "length". Given the read value
> was calculated based on the size of the read buffer, the longer
> response length runs off the end of the buffer.
> 
> [...]

Applied to 5.13/scsi-fixes, thanks!

[1/1] lpfc: Fix bad memory access during VPD DUMP mailbox command
      https://git.kernel.org/mkp/scsi/c/e4ec10228fdf

-- 
Martin K. Petersen	Oracle Linux Engineering