Skip to contentSkip to conversation

ata: libata-scsi: add out-of-bounds checks in ata_scsiop_inq_b9

PATCH New Karuna Ramkumar 2 participants 2 messages Jul 10 libata

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ata: libata-scsi: add out-of-bounds checks in ata_scsiop_inq_b9

The function ata_scsiop_inq_b9() currently attempts to write 64 + nr_cpr * 32 number of bytes. If nr_cpr crosses 62, this can cause a buffer overflow since rbuf is 2048 bytes (ATA_SCSI_RBUF_SIZE) in size.

Fix this by adding bounds check before writing to rbuf. If the total bytes are greater than ATA_SCSI_RBUF_SIZE, then fail the command with ABORTED_COMMAND error and exit the function.

Tested this flow by modifying nr_cpr and confirming that the function exits early, avoiding performing an out-of-bounds write.

Fixes fe22e1c2f705 ("libata: support concurrent positioning ranges log")
Assisted-by Antigravity:gemini-3.5-flash
---
 drivers/ata/libata-scsi.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/ata/libata-scsi.c +8 -1
index 5868526301a2..6f54e23a586c 100644
@@ -2361,15 +2361,22 @@ static unsigned int ata_scsiop_inq_b9(struct ata_device *dev,
 	struct ata_cpr_log *cpr_log = dev->cpr_log;
 	u8 *desc = &rbuf[64];
 	int i;
+	int page_len;
 
 	if (!cpr_log) {
 		ata_scsi_set_invalid_field(dev, cmd, 2, 0xff);
 		return 0;
 	}
 
+	page_len = 64 + (int)cpr_log->nr_cpr * 32;
+	if (page_len > ATA_SCSI_RBUF_SIZE) {
+		ata_scsi_set_sense(dev, cmd, ABORTED_COMMAND, 0, 0);
+		return 0;
+	}
+
 	/* SCSI Concurrent Positioning Ranges VPD page: SBC-5 rev 1 or later */
 	rbuf[1] = 0xb9;
-	put_unaligned_be16(64 + (int)cpr_log->nr_cpr * 32 - 4, &rbuf[2]);
+	put_unaligned_be16(page_len - 4, &rbuf[2]);
 
 	for (i = 0; i < cpr_log->nr_cpr; i++, desc += 32) {
 		desc[0] = cpr_log->cpr[i].num;
Signature · 1 line
2.55.0.795.g602f6c329a-goog
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] The newly added bounds check to prevent `page_len > ATA_SCSI_RBUF_SIZE` is mathematically dead…

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The newly added bounds check to prevent page_len > ATA_SCSI_RBUF_SIZE is mathematically dead code. --

--- Patch [1]: [PATCH] ata: libata-scsi: add out-of-bounds checks in ata_scsiop_inq_b9 --- commit bacad4b494e8c54b7f847f50ed2b764d2bdf3fde
Author: Karuna Ramkumar <[email protected]>
ata: libata-scsi: add out-of-bounds checks in ata_scsiop_inq_b9

This commit adds a bounds check to ata_scsiop_inq_b9() to ensure the calculated page length does not exceed the SCSI request buffer size. If the calculated size is larger, the command is aborted to prevent a buffer overflow.

Quoted 3 lines
drivers/ata/libata-scsi.c +0 -0

[ ... ]

Quoted 1 line
drivers/ata/libata-scsi.c +0 -0
@@ -2361,15 +2361,22 @@ static unsigned int ata_scsiop_inq_b9(struct ata_device *dev,

[ ... ]

drivers/ata/libata-scsi.c static unsigned int ata_scsiop_inq_b9(struct ata_device *dev, · line 2375

Quoted 10 lines
drivers/ata/libata-scsi.c +5 -0
 	if (!cpr_log) {
 		ata_scsi_set_invalid_field(dev, cmd, 2, 0xff);
 		return 0;
 	}
 
+	page_len = 64 + (int)cpr_log->nr_cpr * 32;
+	if (page_len > ATA_SCSI_RBUF_SIZE) {
+		ata_scsi_set_sense(dev, cmd, ABORTED_COMMAND, 0, 0);
+		return 0;
+	}

[Severity: Low]
Is this bounds check reachable during normal operation?

Looking at how cpr_log is populated in ata_read_cpr_log(), nr_cpr is already explicitly validated against ATA_DEV_MAX_CPR. Since ATA_DEV_MAX_CPR is mathematically derived from ATA_SCSI_RBUF_SIZE, a device reporting a value exceeding this limit has its log discarded, leaving dev->cpr_log as NULL.

Because ata_scsiop_inq_b9() immediately returns if cpr_log is NULL right above this snippet, it seems page_len can never evaluate to a value greater than ATA_SCSI_RBUF_SIZE. Does this make the new check mathematically dead code under real inputs?

Quoted 5 lines
drivers/ata/libata-scsi.c +2 -1
+
 	/* SCSI Concurrent Positioning Ranges VPD page: SBC-5 rev 1 or later */
 	rbuf[1] = 0xb9;
-	put_unaligned_be16(64 + (int)cpr_log->nr_cpr * 32 - 4, &rbuf[2]);
+	put_unaligned_be16(page_len - 4, &rbuf[2]);

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1