Skip to contentSkip to conversation

scsi: aacraid: Replace one-element array with flexible-array member

PATCH Applied Gustavo A. R. Silva 4 participants 9 messages Apr 21, 2021 – May 11, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 scsi: aacraid: Replace one-element array with flexible-array member

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

Refactor the code according to the use of a flexible-array member in struct aac_raw_io2 instead of one-element array, and use the struct_size() helper.

Also, this helps with the ongoing efforts to enable -Warray-bounds by fixing the following warnings:

drivers/scsi/aacraid/aachba.c: In function ‘aac_build_sgraw2’:
drivers/scsi/aacraid/aachba.c:3970:18: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]

 3970 |     if (rio2->sge[j].length % (i*PAGE_SIZE)) {
      |         ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:3974:27: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 3974 |     nseg_new += (rio2->sge[j].length / (i*PAGE_SIZE));
      |                  ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4011:28: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4011 |   for (j = 0; j < rio2->sge[i].length / (pages * PAGE_SIZE); ++j) {
      |                   ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4012:24: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4012 |    addr_low = rio2->sge[i].addrLow + j * pages * PAGE_SIZE;
      |               ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4014:33: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4014 |    sge[pos].addrHigh = rio2->sge[i].addrHigh;
      |                        ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4015:28: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4015 |    if (addr_low < rio2->sge[i].addrLow)
      |                   ~~~~~~~~~^~~

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.9/process/deprecated.html#zero-length-and-one-element-arrays

---
Changes in v3:
- Use (nseg_new-1)*sizeof(struct sge_ieee1212) to calculate size in call to memcpy() in order to avoid any confusion.

Changes in v2:
- Add code comment for clarification.

 drivers/scsi/aacraid/aachba.c  | 10 +++++-----
 drivers/scsi/aacraid/aacraid.h |  2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)
drivers/scsi/aacraid/aachba.c +5 -5
index f1f62b5da8b7..46b8dffce2dd 100644
@@ -1235,8 +1235,8 @@ static int aac_read_raw_io(struct fib * fib, struct scsi_cmnd * cmd, u64 lba, u3
 		if (ret < 0)
 			return ret;
 		command = ContainerRawIo2;
-		fibsize = sizeof(struct aac_raw_io2) +
-			((le32_to_cpu(readcmd2->sgeCnt)-1) * sizeof(struct sge_ieee1212));
+		fibsize = struct_size(readcmd2, sge,
+				     le32_to_cpu(readcmd2->sgeCnt));
 	} else {
 		struct aac_raw_io *readcmd;
 		readcmd = (struct aac_raw_io *) fib_data(fib);
@@ -1366,8 +1366,8 @@ static int aac_write_raw_io(struct fib * fib, struct scsi_cmnd * cmd, u64 lba, u
 		if (ret < 0)
 			return ret;
 		command = ContainerRawIo2;
-		fibsize = sizeof(struct aac_raw_io2) +
-			((le32_to_cpu(writecmd2->sgeCnt)-1) * sizeof(struct sge_ieee1212));
+		fibsize = struct_size(writecmd2, sge,
+				      le32_to_cpu(writecmd2->sgeCnt));
 	} else {
 		struct aac_raw_io *writecmd;
 		writecmd = (struct aac_raw_io *) fib_data(fib);
@@ -3998,7 +3998,7 @@ static int aac_convert_sgraw2(struct aac_raw_io2 *rio2, int pages, int nseg, int
 	if (aac_convert_sgl == 0)
 		return 0;
 
-	sge = kmalloc_array(nseg_new, sizeof(struct sge_ieee1212), GFP_ATOMIC);
+	sge = kmalloc_array(nseg_new, sizeof(*sge), GFP_ATOMIC);
 	if (sge == NULL)
 		return -ENOMEM;
 
drivers/scsi/aacraid/aacraid.h +1 -1
index e3e4ecbea726..3733df77bc65 100644
@@ -1929,7 +1929,7 @@ struct aac_raw_io2 {
 	u8		bpComplete;	/* reserved for F/W use */
 	u8		sgeFirstIndex;	/* reserved for F/W use */
 	u8		unused[4];
-	struct sge_ieee1212	sge[1];
+	struct sge_ieee1212	sge[];
 };
 
 #define CT_FLUSH_CACHE 129
Signature · 1 line
2.27.0
#1 lore raw

Hi Gustavo!

Quoted 3 lines

Changes in v3:
- Use (nseg_new-1)*sizeof(struct sge_ieee1212) to calculate size in call to memcpy() in order to avoid any confusion.

The amended memcpy() hunk appears to be missing from the v3 patch.

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#2 lore raw
Quoted 8 lines · Martin K. Petersen

Hi Gustavo!

Quoted 3 lines

Changes in v3:
- Use (nseg_new-1)*sizeof(struct sge_ieee1212) to calculate size in call to memcpy() in order to avoid any confusion.

The amended memcpy() hunk appears to be missing from the v3 patch.

It's unchanged from the perspective of the original code. (i.e. there's no need to change it since that memcpy isn't involved in anything changed by the swapping to the flexible array.)

Signature · 1 line
Kees Cook
#3 lore raw

Kees,

Quoted 5 lines
Quoted 1 line

The amended memcpy() hunk appears to be missing from the v3 patch.

It's unchanged from the perspective of the original code. (i.e. there's no need to change it since that memcpy isn't involved in anything changed by the swapping to the flexible array.)

Ah, I was under the impression that you intended to do sizeof(*sge) to match the kmalloc_array() above.

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#4 lore raw
In reply to Gustavo A. R. Silva

Hi Martin,

Friendly ping: could you take this patch, please? :)

Thanks!
--
Gustavo

Quoted 101 lines · Gustavo A. R. Silva

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

Refactor the code according to the use of a flexible-array member in struct aac_raw_io2 instead of one-element array, and use the struct_size() helper.

Also, this helps with the ongoing efforts to enable -Warray-bounds by fixing the following warnings:

drivers/scsi/aacraid/aachba.c: In function ‘aac_build_sgraw2’:
drivers/scsi/aacraid/aachba.c:3970:18: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]

 3970 |     if (rio2->sge[j].length % (i*PAGE_SIZE)) {
      |         ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:3974:27: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 3974 |     nseg_new += (rio2->sge[j].length / (i*PAGE_SIZE));
      |                  ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4011:28: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4011 |   for (j = 0; j < rio2->sge[i].length / (pages * PAGE_SIZE); ++j) {
      |                   ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4012:24: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4012 |    addr_low = rio2->sge[i].addrLow + j * pages * PAGE_SIZE;
      |               ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4014:33: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4014 |    sge[pos].addrHigh = rio2->sge[i].addrHigh;
      |                        ~~~~~~~~~^~~
drivers/scsi/aacraid/aachba.c:4015:28: warning: array subscript 1 is above array bounds of ‘struct sge_ieee1212[1]’ [-Warray-bounds]
 4015 |    if (addr_low < rio2->sge[i].addrLow)
      |                   ~~~~~~~~~^~~

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.9/process/deprecated.html#zero-length-and-one-element-arrays

---
Changes in v3:
- Use (nseg_new-1)*sizeof(struct sge_ieee1212) to calculate size in call to memcpy() in order to avoid any confusion.

Changes in v2:
- Add code comment for clarification.

 drivers/scsi/aacraid/aachba.c  | 10 +++++-----
 drivers/scsi/aacraid/aacraid.h |  2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)
drivers/scsi/aacraid/aachba.c +5 -5
index f1f62b5da8b7..46b8dffce2dd 100644
@@ -1235,8 +1235,8 @@ static int aac_read_raw_io(struct fib * fib, struct scsi_cmnd * cmd, u64 lba, u3
 		if (ret < 0)
 			return ret;
 		command = ContainerRawIo2;
-		fibsize = sizeof(struct aac_raw_io2) +
-			((le32_to_cpu(readcmd2->sgeCnt)-1) * sizeof(struct sge_ieee1212));
+		fibsize = struct_size(readcmd2, sge,
+				     le32_to_cpu(readcmd2->sgeCnt));
 	} else {
 		struct aac_raw_io *readcmd;
 		readcmd = (struct aac_raw_io *) fib_data(fib);
@@ -1366,8 +1366,8 @@ static int aac_write_raw_io(struct fib * fib, struct scsi_cmnd * cmd, u64 lba, u
 		if (ret < 0)
 			return ret;
 		command = ContainerRawIo2;
-		fibsize = sizeof(struct aac_raw_io2) +
-			((le32_to_cpu(writecmd2->sgeCnt)-1) * sizeof(struct sge_ieee1212));
+		fibsize = struct_size(writecmd2, sge,
+				      le32_to_cpu(writecmd2->sgeCnt));
 	} else {
 		struct aac_raw_io *writecmd;
 		writecmd = (struct aac_raw_io *) fib_data(fib);
@@ -3998,7 +3998,7 @@ static int aac_convert_sgraw2(struct aac_raw_io2 *rio2, int pages, int nseg, int
 	if (aac_convert_sgl == 0)
 		return 0;
 
-	sge = kmalloc_array(nseg_new, sizeof(struct sge_ieee1212), GFP_ATOMIC);
+	sge = kmalloc_array(nseg_new, sizeof(*sge), GFP_ATOMIC);
 	if (sge == NULL)
 		return -ENOMEM;
 
drivers/scsi/aacraid/aacraid.h +1 -1
index e3e4ecbea726..3733df77bc65 100644
@@ -1929,7 +1929,7 @@ struct aac_raw_io2 {
 	u8		bpComplete;	/* reserved for F/W use */
 	u8		sgeFirstIndex;	/* reserved for F/W use */
 	u8		unused[4];
-	struct sge_ieee1212	sge[1];
+	struct sge_ieee1212	sge[];
 };
 
 #define CT_FLUSH_CACHE 129
#5 lore raw

Gustavo,

Quoted 1 line

Friendly ping: could you take this patch, please? :)

Applied to 5.14/scsi-staging, thanks!

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#6 lore raw
Quoted 1 line · Martin K. Petersen

Applied to 5.14/scsi-staging, thanks!

Awesome! :)

Thank you.
--
Gustavo

#7 lore raw
In reply to Gustavo A. R. Silva
Quoted 10 lines · Gustavo A. R. Silva

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

Refactor the code according to the use of a flexible-array member in struct aac_raw_io2 instead of one-element array, and use the struct_size() helper.

[...]

Applied to 5.14/scsi-queue, thanks!

[1/1] scsi: aacraid: Replace one-element array with flexible-array member

      https://git.kernel.org/mkp/scsi/c/39107e8577ad
Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#8 lore raw
Quoted 11 lines · Martin K. Petersen
Quoted 9 lines · Gustavo A. R. Silva

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

Refactor the code according to the use of a flexible-array member in struct aac_raw_io2 instead of one-element array, and use the struct_size() helper.

Quoted 6 lines
Quoted 1 line

[...]

Applied to 5.14/scsi-queue, thanks!

[1/1] scsi: aacraid: Replace one-element array with flexible-array member

      https://git.kernel.org/mkp/scsi/c/39107e8577ad

Awesome. :)

Thanks, Martin.
--
Gustavo