Skip to contentSkip to conversation

raw/ioat: fix queue index calculation

PATCH Applied Bruce Richardson 2 participants 6 messages Nov 3, 2020 – Nov 13, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 raw/ioat: fix queue index calculation

Coverity flags a possible problem where the 8-bit wq_idx value may have errors when shifted and sign-extended to 64-bits. Since this can only occur if the shift index is larger than any expected value from hardware, it's unlikely to cause any real problems, but we can eliminate any possible errors, and the coverity issue, by explicitly typecasting the uint8_t value to 64-bits before any shift operations occur.

Coverity issue: 363695

Fixes a33969462135 ("raw/ioat: fix work-queue config size")
---
 drivers/raw/ioat/idxd_pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/raw/ioat/idxd_pci.c +2 -1
index 99ecfbbb4..7bc125014 100644
@@ -47,7 +47,8 @@ idxd_pci_dev_command(struct idxd_rawdev *idxd, enum rte_idxd_cmds command)
 static uint32_t *
 idxd_get_wq_cfg(struct idxd_pci_common *pci, uint8_t wq_idx)
 {
-	return RTE_PTR_ADD(pci->wq_regs_base, wq_idx << (5 + pci->wq_cfg_sz));
+	return RTE_PTR_ADD(pci->wq_regs_base,
+			(uint64_t)wq_idx << (5 + pci->wq_cfg_sz));
 }
 
 static int
Signature · 1 line
2.25.1
#1 lore raw

03/11/2020 15:48, Bruce Richardson:

Quoted 11 lines

Coverity flags a possible problem where the 8-bit wq_idx value may have errors when shifted and sign-extended to 64-bits. Since this can only occur if the shift index is larger than any expected value from hardware, it's unlikely to cause any real problems, but we can eliminate any possible errors, and the coverity issue, by explicitly typecasting the uint8_t value to 64-bits before any shift operations occur.

Coverity issue: 363695

Fixes a33969462135 ("raw/ioat: fix work-queue config size")

Applied, thanks

#2 lore raw

03/11/2020 21:21, Thomas Monjalon:

Quoted 8 lines

03/11/2020 15:48, Bruce Richardson:

Quoted 7 lines

Coverity flags a possible problem where the 8-bit wq_idx value may have errors when shifted and sign-extended to 64-bits. Since this can only occur if the shift index is larger than any expected value from hardware, it's unlikely to cause any real problems, but we can eliminate any possible errors, and the coverity issue, by explicitly typecasting the uint8_t value to 64-bits before any shift operations occur.

Quoted 6 lines
Quoted 4 lines

Coverity issue: 363695

Fixes a33969462135 ("raw/ioat: fix work-queue config size")

Applied, thanks

Actually no, sorry for the confusion, there is a build issue with 32-bit compiler:
error: cast to pointer from integer of different size

(thanks David for the heads up)

#3 lore raw
Quoted 14 lines · Thomas Monjalon

03/11/2020 21:21, Thomas Monjalon:

Quoted 13 lines

03/11/2020 15:48, Bruce Richardson:

Quoted 11 lines

Coverity flags a possible problem where the 8-bit wq_idx value may have errors when shifted and sign-extended to 64-bits. Since this can only occur if the shift index is larger than any expected value from hardware, it's unlikely to cause any real problems, but we can eliminate any possible errors, and the coverity issue, by explicitly typecasting the uint8_t value to 64-bits before any shift operations occur.

Coverity issue: 363695

Fixes a33969462135 ("raw/ioat: fix work-queue config size")
Quoted 8 lines
Quoted 1 line

Applied, thanks

Actually no, sorry for the confusion, there is a build issue with 32-bit compiler:
error: cast to pointer from integer of different size

(thanks David for the heads up)

Ok, thanks for heads-up, I'll see about a v2.

#4 lore raw

PATCH v2 1/1 raw/ioat: fix queue index calculation

In reply to Bruce Richardson

Coverity flags a possible problem where the 8-bit wq_idx value may have errors when shifted and sign-extended to pointer size. Since this can only occur if the shift index is larger than any expected value from hardware, it's unlikely to cause any real problems, but we can eliminate any possible errors, and the coverity issue, by explicitly typecasting the uint8_t value to uintptr_t before any shift operations occur.

Coverity issue: 363695

Fixes a33969462135 ("raw/ioat: fix work-queue config size")

---
V2: Change uint64_t cast to uintptr_t to prevent 32-bit build issues

---
 drivers/raw/ioat/idxd_pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/raw/ioat/idxd_pci.c +2 -1
index 99ecfbbb4..01623f33f 100644
@@ -47,7 +47,8 @@ idxd_pci_dev_command(struct idxd_rawdev *idxd, enum rte_idxd_cmds command)
 static uint32_t *
 idxd_get_wq_cfg(struct idxd_pci_common *pci, uint8_t wq_idx)
 {
-	return RTE_PTR_ADD(pci->wq_regs_base, wq_idx << (5 + pci->wq_cfg_sz));
+	return RTE_PTR_ADD(pci->wq_regs_base,
+			(uintptr_t)wq_idx << (5 + pci->wq_cfg_sz));
 }
 
 static int
Signature · 1 line
2.25.1
#5 lore raw

05/11/2020 11:41, Bruce Richardson:

Quoted 11 lines

Coverity flags a possible problem where the 8-bit wq_idx value may have errors when shifted and sign-extended to pointer size. Since this can only occur if the shift index is larger than any expected value from hardware, it's unlikely to cause any real problems, but we can eliminate any possible errors, and the coverity issue, by explicitly typecasting the uint8_t value to uintptr_t before any shift operations occur.

Coverity issue: 363695

Fixes a33969462135 ("raw/ioat: fix work-queue config size")

Applied, thanks