pcmcia: synclink_cs: replace 1-element array with flex-array member
PATCH 1/1 pcmcia: synclink_cs: replace 1-element array with flex-array member
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly.
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the FORTIFY_SOURCE routines on memcpy() and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].
--- drivers/char/pcmcia/synclink_cs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/char/pcmcia/synclink_cs.c +3 -2
@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */
@@ -2611,7 +2611,8 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; + info->rx_buf_size =max (offsetof(typeof (RXBUF ), data) +1 ,sizeof (RXBUF )) + + info->max_frame_size;/* calculate total allocation size for 8 buffers */ info->rx_buf_total_size = info->rx_buf_size *8 ;
Signature · 1 line
2.38.1
PATCH v2 1/1 pcmcia: synclink_cs: replace 1-element array with flex-array member
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF. No changes were required within the source code because of the existing padding in RXBUF struct
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the FORTIFY_SOURCE routines on memcpy() and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].
---
Changelog:
- v2: removed changes to how the size of RXBUF was calculated. I changed my mind after thinking about the existing padding in the struct. Happy to discuss it if anyone sees it differently.
- v1: https://lore.kernel.org/lkml/[email protected]/
--- drivers/char/pcmcia/synclink_cs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/char/pcmcia/synclink_cs.c +1 -1
@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */
Signature · 1 line
2.38.1
Quoted 5 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF. No changes were required within the source code because of the existing padding in RXBUF struct
You shouldn't rely on padding. Make you change robust independently on the padding. See also below.
Quoted 2 lines
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This is interesting...
Quoted 8 lines
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].
The blank lines are not allowed in the tag block (in case you want to have Link: to be recognized as a tag).
Quoted 7 lines
Signed-off-by Paulo Miguel Almeida <[email protected]>---
Changelog:- v2: removed changes to how the size of RXBUF was calculated. I changed my mind after thinking about the existing padding in the struct. Happy to discuss it if anyone sees it differently.
I feel worried about in particular this code:
/* each buffer has header and data */ info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size;
which means that entire rx_alloc_buffers() should be revisited. Also take into account the use of one or more macros from overflow.h for memory allocation.
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 36 lines · Andy Shevchenko
Quoted 5 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF. No changes were required within the source code because of the existing padding in RXBUF struct
You shouldn't rely on padding. Make you change robust independently on the padding. See also below.
Quoted 2 lines
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This is interesting...
Quoted 8 lines
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].The blank lines are not allowed in the tag block (in case you want to have Link: to be recognized as a tag).
Quoted 7 lines
Signed-off-by Paulo Miguel Almeida <[email protected]>---
Changelog:- v2: removed changes to how the size of RXBUF was calculated. I changed my mind after thinking about the existing padding in the struct. Happy to discuss it if anyone sees it differently.
Quoted 12 lines
I feel worried about in particular this code:
/* each buffer has header and data */
info->rx_buf_size= sizeof(RXBUF) +info->max_frame_size;which means that entire
rx_alloc_buffers()should be revisited. Also take into account the use of one or more macros fromoverflow.hfor memory allocation.Signature · 2 lines
With Best Regards, Andy Shevchenko
Hi Kees, Hi Andy, Thanks for taking the time to review this patch.
As both of you had similar points, I will reply them here.
The reasons why it had no binary changes was because of the combination of this 2 things:
1) Existing padding - so sizeof(RXBUF) returned 8 bytes in both cases.
pahole -C RXBUF gcc/before/drivers/char/pcmcia/synclink_cs.ko typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[1]; /* 5 1 */ /* size: 8, cachelines: 1, members: 3 */
/* padding: 2 */
/* last cacheline: 8 bytes */
} RXBUF;pahole -C RXBUF gcc/after/drivers/char/pcmcia/synclink_cs.ko
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[]; /* 5 0 */
/* size: 8, cachelines: 1, members: 3 */
/* padding: 3 */
/* last cacheline: 8 bytes */
} RXBUF;2) RXBUF (as implemented now) is just like a pair of lenses from which a developer can have access to one of the circular buffers in MGSLPC_INFO struct called 'rx_buf'.
2611 static int rx_alloc_buffers(MGSLPC_INFO *info) 2612 {
2613 /* each buffer has header and data */
2614 info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size;
2615
2616 /* calculate total allocation size for 8 buffers */
2617 info->rx_buf_total_size = info->rx_buf_size * 8;
2618
2619 /* limit total allocated memory */
2620 if (info->rx_buf_total_size > 0x10000)
2621 info->rx_buf_total_size = 0x10000;
2622
2623 /* calculate number of buffers */
2624 info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size;
2625
2626 info->rx_buf = kmalloc(info->rx_buf_total_size, GFP_KERNEL);
To be honest, char data[_1_] in RXBUF was never required to be there. The code base seems to make sure that it doesn't run past its limits by keeping track of size buffer on MGSLPC_INFO->rx_buf_size (and sometimes RXBUF->count)
(Addressing one point made by Andy about using of of the macros in overflow.h)
struct_size(buf, data, 1) would return 9 bytes which could potentially break the existing driver as it produces binary changes.
Let me know your thoughts
thanks!
- Paulo A.
Quoted 34 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly.
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].Signed-off-by Paulo Miguel Almeida <[email protected]>--- drivers/char/pcmcia/synclink_cs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)drivers/char/pcmcia/synclink_cs.c +1
-1@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */
drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2615
Quoted 6 lines
drivers/char/pcmcia/synclink_cs.c +2
-1@@ -2611,7 +2611,8 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; + info->rx_buf_size =max (offsetof(typeof (RXBUF ), data) +1 ,sizeof (RXBUF )) + + info->max_frame_size;
It seems like there is an existing size bug here, and likely should be fixed separately?
i.e. this was already allocating 1 byte "too much". I'd expect this first:
- info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size;
+ info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;
and then the next patch:
- char data[1];
+ char data[];
...
- info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;
+ info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size;
The above would induce a binary output change, and the second would not.
Though this results in what you had for the v2 patch (but I can't believe it had no binary changes...)
Signature · 1 line
Kees Cook
Quoted 59 lines · Kees Cook
Quoted 34 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly.
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].Signed-off-by Paulo Miguel Almeida <[email protected]>--- drivers/char/pcmcia/synclink_cs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)drivers/char/pcmcia/synclink_cs.c +1
-1@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */ drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2615
Quoted 6 lines
drivers/char/pcmcia/synclink_cs.c +2
-1@@ -2611,7 +2611,8 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; + info->rx_buf_size =max (offsetof(typeof (RXBUF ), data) +1 ,sizeof (RXBUF )) + + info->max_frame_size;It seems like there is an existing size bug here, and likely should be fixed separately?
i.e. this was already allocating 1 byte "too much". I'd expect this first:
(quoted diff) +1
-1- info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size; + info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;and then the next patch:
drivers/char/pcmcia/synclink_cs.c +2
-2-char data[1 ]; +char data[]; - info->rx_buf_size =sizeof (RXBUF ) -1 + info->max_frame_size; + info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size;
Quoted 7 lines
The above would induce a binary output change, and the second would not.
Though this results in what you had for the v2 patch (but I can't believe it had no binary changes...)
Signature · 1 line
Kees Cook
Hi Kees, Hi Andy, Thanks for taking the time to review this patch.
As both of you had similar points, I will reply them here.
The reasons why it had no binary changes was because of the combination of this 2 things:
1) Existing padding - so sizeof(RXBUF) returned 8 bytes in both cases.
pahole -C RXBUF gcc/before/drivers/char/pcmcia/synclink_cs.ko
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[1]; /* 5 1 */
/* size: 8, cachelines: 1, members: 3 */
/* padding: 2 */
/* last cacheline: 8 bytes */
} RXBUF;pahole -C RXBUF gcc/after/drivers/char/pcmcia/synclink_cs.ko
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[]; /* 5 0 */
/* size: 8, cachelines: 1, members: 3 */
/* padding: 3 */
/* last cacheline: 8 bytes */
} RXBUF;2) RXBUF (as implemented now) is just like a pair of lenses from which a developer can have access to one of the circular buffers in MGSLPC_INFO struct called 'rx_buf'.
2611 static int rx_alloc_buffers(MGSLPC_INFO *info) 2612 {
2613 /* each buffer has header and data */
2614 info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size;
2615
2616 /* calculate total allocation size for 8 buffers */
2617 info->rx_buf_total_size = info->rx_buf_size * 8;
2618
2619 /* limit total allocated memory */
2620 if (info->rx_buf_total_size > 0x10000)
2621 info->rx_buf_total_size = 0x10000;
2622
2623 /* calculate number of buffers */
2624 info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size;
2625
2626 info->rx_buf = kmalloc(info->rx_buf_total_size, GFP_KERNEL);
To be honest, char data[_1_] in RXBUF was never required to be there. The code base seems to make sure that it doesn't run past its limits by keeping track of size buffer on MGSLPC_INFO->rx_buf_size (and sometimes RXBUF->count)
(Addressing one point made by Andy about using of of the macros in overflow.h)
struct_size(buf, data, 1) would return 9 bytes which could potentially break the existing driver as it produces binary changes.
Let me know your thoughts
thanks!
- Paulo A.
Quoted 89 lines · Paulo Miguel Almeida
Quoted 59 lines · Kees Cook
Quoted 34 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly.
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].Signed-off-by Paulo Miguel Almeida <[email protected]>--- drivers/char/pcmcia/synclink_cs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)drivers/char/pcmcia/synclink_cs.c +1
-1@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */ drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2615
Quoted 6 lines
drivers/char/pcmcia/synclink_cs.c +2
-1@@ -2611,7 +2611,8 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; + info->rx_buf_size =max (offsetof(typeof (RXBUF ), data) +1 ,sizeof (RXBUF )) + + info->max_frame_size;It seems like there is an existing size bug here, and likely should be fixed separately?
i.e. this was already allocating 1 byte "too much". I'd expect this first:
(quoted diff) +1
-1- info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size; + info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;and then the next patch:
drivers/char/pcmcia/synclink_cs.c +2
-2-char data[1 ]; +char data[]; - info->rx_buf_size =sizeof (RXBUF ) -1 + info->max_frame_size; + info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size;Quoted 7 lines
The above would induce a binary output change, and the second would not.
Though this results in what you had for the v2 patch (but I can't believe it had no binary changes...)
Signature · 1 line
Kees CookHi Kees, Hi Andy, Thanks for taking the time to review this patch.
As both of you had similar points, I will reply them here.
The reasons why it had no binary changes was because of the combination of this 2 things:
1) Existing padding - so sizeof(RXBUF) returned 8 bytes in both cases.
pahole -C RXBUF
gcc/before/drivers/char/pcmcia/synclink_cs.kotypedef struct { int count; /* 0 4 */ unsigned char status; /* 4 1 */ char data[1]; /* 5 1 */ /* size: 8, cachelines: 1, members: 3 */ /* padding: 2 */ /* last cacheline: 8 bytes */ } RXBUF;pahole -C RXBUF
gcc/after/drivers/char/pcmcia/synclink_cs.ko
Quoted 9 lines
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[]; /* 5 0 */
/* size: 8, cachelines: 1, members: 3 */
/* padding: 3 */
/* last cacheline: 8 bytes */
} RXBUF;Ah-ha, now I see.
Quoted 27 lines
2) RXBUF (as implemented now) is just like a pair of lenses from which a developer can have access to one of the circular buffers in
MGSLPC_INFOstruct called 'rx_buf'.2611 static int
rx_alloc_buffers(MGSLPC_INFO *info)2612 {
2613 /* each buffer has header and data */
2614info->rx_buf_size= sizeof(RXBUF) +info->max_frame_size;
2615
2616 /* calculate total allocation size for 8 buffers */
2617info->rx_buf_total_size=info->rx_buf_size* 8;
2618
2619 /* limit total allocated memory */
2620 if (info->rx_buf_total_size>0x10000)
2621info->rx_buf_total_size=0x10000;
2622
2623 /* calculate number of buffers */
2624info->rx_buf_count=info->rx_buf_total_size/info->rx_buf_size;
2625
2626info->rx_buf=kmalloc(info->rx_buf_total_size, GFP_KERNEL);To be honest, char data[_1_] in RXBUF was never required to be there. The code base seems to make sure that it doesn't run past its limits by keeping track of size buffer on
MGSLPC_INFO->rx_buf_size(and sometimesRXBUF->count)
Quoted 5 lines
(Addressing one point made by Andy about using of of the macros in
overflow.h)
struct_size(buf, data, 1)would return 9 bytes which could potentially break the existing driver as it produces binary changes.
Yeah, I think your v2 is fine. Perhaps explicitly repeat the notes about struct size padding in a v3 commit log?
Signature · 1 line
Kees Cook
Quoted 13 lines · Paulo Miguel Almeida
Quoted 12 lines · Kees Cook
Quoted 11 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly.
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].
...
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
typedef struct { int count; unsigned char status;drivers/char/pcmcia/synclink_cs.c +1
-1-char data[1 ]; +char data[]; }RXBUF ;
...
Quoted 19 lines
As both of you had similar points, I will reply them here.
The reasons why it had no binary changes was because of the combination of this 2 things:
1) Existing padding - so sizeof(RXBUF) returned 8 bytes in both cases.
pahole -C RXBUF
gcc/before/drivers/char/pcmcia/synclink_cs.kotypedef struct {int count; /* 0 4 */ unsigned char status; /* 4 1 */ char data[1]; /* 5 1 *//* size: 8, cachelines: 1, members: 3 */ /* padding: 2 */ /* last cacheline: 8 bytes */ } RXBUF;pahole -C RXBUF
gcc/after/drivers/char/pcmcia/synclink_cs.ko
Quoted 9 lines
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[]; /* 5 0 */
/* size: 8, cachelines: 1, members: 3 */
/* padding: 3 */
/* last cacheline: 8 bytes */
} RXBUF;Yes, and Try to make it work with __packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.
Quoted 26 lines
2) RXBUF (as implemented now) is just like a pair of lenses from which a developer can have access to one of the circular buffers in
MGSLPC_INFOstruct called 'rx_buf'.2611 static int
rx_alloc_buffers(MGSLPC_INFO *info)2612 {
2613 /* each buffer has header and data */
2614info->rx_buf_size= sizeof(RXBUF) +info->max_frame_size;
2615
2616 /* calculate total allocation size for 8 buffers */
2617info->rx_buf_total_size=info->rx_buf_size* 8;
2618
2619 /* limit total allocated memory */
2620 if (info->rx_buf_total_size>0x10000)
2621info->rx_buf_total_size=0x10000;
2622
2623 /* calculate number of buffers */
2624info->rx_buf_count=info->rx_buf_total_size/info->rx_buf_size;
2625
2626info->rx_buf=kmalloc(info->rx_buf_total_size, GFP_KERNEL);To be honest, char data[_1_] in RXBUF was never required to be there. The code base seems to make sure that it doesn't run past its limits by keeping track of size buffer on
MGSLPC_INFO->rx_buf_size(and sometimesRXBUF->count)
Quoted 5 lines
(Addressing one point made by Andy about using of of the macros in
overflow.h)
struct_size(buf, data, 1)would return 9 bytes which could potentially break the existing driver as it produces binary changes.
You got it incorrectly. I believe you should use something different than 1. In previous lines in the function it multiplies sizeof + max_frame_size by 8.
The full change should be something like
check_add(sizeof(), max_frame_size)kcalloc(8, size)
Think about it.
Quoted 1 line
Let me know your thoughts
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 4 lines · Andy Shevchenko
Yes, and Try to make it work with
__packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.
I meant that v2 is functionally identical to the existing code.
Quoted 4 lines
The full change should be something like
check_add(sizeof(), max_frame_size)kcalloc(8, size)
Right -- this would fix the existing mistakes in size calculation (and is certainly better).
Signature · 1 line
Kees Cook
Quoted 7 lines · Kees Cook
Quoted 4 lines · Andy Shevchenko
Yes, and Try to make it work with
__packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.I meant that v2 is functionally identical to the existing code.
Ah, sorry for misunderstanding.
Quoted 7 lines
Quoted 4 lines
The full change should be something like
check_add(sizeof(), max_frame_size)kcalloc(8, size)Right -- this would fix the existing mistakes in size calculation (and is certainly better).
Glad to hear that we are on the same page.
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 11 lines · Andy Shevchenko
Quoted 7 lines · Kees Cook
Quoted 4 lines · Andy Shevchenko
Yes, and Try to make it work with
__packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.I meant that v2 is functionally identical to the existing code.
Ah, sorry for misunderstanding.
I agree with using __packed attribute to remove the extra padding (and for the reasons you mentioned before). That would reduce the sizeof(RXBUF) from 8 to 5 (which is good) but that is still 1 byte "too much".
Piggying back on a suggestion Kees gave before:
- info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size;
+ info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;
That way RXBUF->data will point to the first byte of the frame_size (MGSLPC_INFO->max_frame_size) which is what is actually needed.
Quoted 10 lines
Quoted 7 lines
Quoted 4 lines
The full change should be something like
check_add(sizeof(), max_frame_size)kcalloc(8, size)Right -- this would fix the existing mistakes in size calculation (and is certainly better).
Glad to hear that we are on the same page.
That makes sense to me.
thanks!
- Paulo A.
Quoted 22 lines · Paulo Miguel Almeida
Quoted 11 lines · Andy Shevchenko
Quoted 7 lines · Kees Cook
Quoted 4 lines · Andy Shevchenko
Yes, and Try to make it work with
__packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.I meant that v2 is functionally identical to the existing code.
Ah, sorry for misunderstanding.
I agree with using
__packedattribute to remove the extra padding (and for the reasons you mentioned before). That would reduce the sizeof(RXBUF) from 8 to 5 (which is good) but that is still 1 byte "too much".Piggying back on a suggestion Kees gave before:
(quoted diff) +1
-1- info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size; + info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;
Quoted 3 lines
That way
RXBUF->datawill point to the first byte of theframe_size(MGSLPC_INFO->max_frame_size) which is what is actually needed.
I chose my words poorly here... sorry my brain is a bit fried today.
Let me rephrase that last sentence. After that change (or similar change), RXBUF->data will point to the first byte of the buffer allocated during the initialisation process. (which is limited/controlled by the size of MGSLPC_INFO->max_frame_size)... so no 'extra byte/padding' will be there.
- Paulo A.
Quoted 16 lines
Quoted 10 lines
Quoted 7 lines
Quoted 4 lines
The full change should be something like
check_add(sizeof(), max_frame_size)kcalloc(8, size)Right -- this would fix the existing mistakes in size calculation (and is certainly better).
Glad to hear that we are on the same page.
That makes sense to me.
thanks!
- Paulo A.
Quoted 3 lines · Paulo Miguel Almeida
Quoted 2 lines · Andy Shevchenko
Quoted 1 line · Kees Cook
On Wed, Dec 14, 2022 at 10:39:52PM +0200, Andy Shevchenko wrote:
...
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 4 lines
Yes, and Try to make it work with
__packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
I meant that v2 is functionally identical to the existing code.
Ah, sorry for misunderstanding.
I agree with using
__packedattribute to remove the extra padding (and for the reasons you mentioned before). That would reduce the sizeof(RXBUF) from 8 to 5 (which is good) but that is still 1 byte "too much".
What I meant with the above is that the code has to work properly with or without __packed. It's just to show you that this code has flaws if it relies on the padding.
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 17 lines · Andy Shevchenko
Quoted 3 lines · Paulo Miguel Almeida
Quoted 2 lines · Andy Shevchenko
Quoted 1 line · Kees Cook
On Wed, Dec 14, 2022 at 10:39:52PM +0200, Andy Shevchenko wrote:
...
Quoted 9 lines
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Yes, and Try to make it work with
__packed. As I said, the problem is that the code is relying on something which is architecture dependent strictly speaking. And hence I disagree with Kees that v2 is okay to go.I meant that v2 is functionally identical to the existing code.
Ah, sorry for misunderstanding.
Quoted 8 lines
Quoted 3 lines
I agree with using
__packedattribute to remove the extra padding (and for the reasons you mentioned before). That would reduce the sizeof(RXBUF) from 8 to 5 (which is good) but that is still 1 byte "too much".What I meant with the above is that the code has to work properly with or without
__packed. It's just to show you that this code has flaws if it relies on the padding.
Right - that would work just as well. I will work on v3 with the suggestions given by you (sizing calculation amendments using overflow.h macros) and kees (adding the notes regarding the padding) then.
- Paulo A.
PATCH v3 1/1 pcmcia: synclink_cs: replace 1-element array with flex-array member
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly. While at it, fix an edge case which could cause rx_buf_count to be 0 when max_frame_size was set to the maximum allowed value (65535).
It's worth mentioning that struct RXBUF was allocating 1 byte "too much" for what is required (ignoring bytes added by padding).
This helps with the ongoing efforts to tighten the FORTIFY_SOURCE routines on memcpy() and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].
---
Changelog:
- v3:
fix size calculation mistakes using overflow.h macros: (Req: Andy Shevchenko, Kees Cook)
add notes struct RXBUF size (Kees Cook)
- v2: removed changes to how the size of RXBUF was calculated. I changed my mind after thinking about the existing padding in the struct. Happy to discuss it if anyone sees it differently.
- v1: https://lore.kernel.org/lkml/[email protected]/
--- drivers/char/pcmcia/synclink_cs.c | 33 +++++++++++++++++++------------ 1 file changed, 20 insertions(+), 13 deletions(-)
drivers/char/pcmcia/synclink_cs.c +20 -13
@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */
@@ -229,12 +229,18 @@ typedef struct _mgslpc_info { }MGSLPC_INFO ;#define MGSLPC_MAGIC 0x5402 +#define MGSLPC_MAX_FRAME_SIZE 65535 +#define MGSLPC_MIN_FRAME_SIZE 4096/* * The size of the serial xmit buffer is 1 page, or 4096 bytes */ #define TXBUFSIZE 4096 +/* +* RXBUF accommodates at least 1 buffer (header+data) of MGSLPC_MAX_FRAME_SIZE +*/ +#define RXBUF_MAX_SIZE (sizeof(RXBUF) + MGSLPC_MAX_FRAME_SIZE)#define CHA 0x00/* channel A offset */ #define CHB 0x40/* channel B offset */
@@ -529,7 +535,7 @@ static int mgslpc_probe(struct pcmcia_device *link)tty_port_init (&info->port); info->port.ops = &mgslpc_port_ops;INIT_WORK (&info->task, bh_handler); - info->max_frame_size =4096 ; + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ;init_waitqueue_head (&info->status_event_wait_q);init_waitqueue_head (&info->event_wait_q);spin_lock_init (&info->lock);
@@ -2611,19 +2617,20 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; +if (check_add_overflow (sizeof (RXBUF ), info->max_frame_size, &info->rx_buf_size)) +return -EINVAL ; -/* calculate total allocation size for 8 buffers */ - info->rx_buf_total_size = info->rx_buf_size *8 ; +/* try to alloc as many buffers that can fit within RXBUF_MAX_SIZE (up to 8) */ +if (check_mul_overflow (info->rx_buf_size,8 , &info->rx_buf_total_size)) +return -EINVAL ; -/* limit total allocated memory */ -if (info->rx_buf_total_size >0x10000 ) - info->rx_buf_total_size =0x10000 ; +if (info->rx_buf_total_size >RXBUF_MAX_SIZE ) + info->rx_buf_total_size =RXBUF_MAX_SIZE ;/* calculate number of buffers */ info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size; - info->rx_buf =kmalloc (info->rx_buf_total_size,GFP_KERNEL ); + info->rx_buf =kcalloc (info->rx_buf_count, info->rx_buf_size,GFP_KERNEL );if (info->rx_buf ==NULL )return -ENOMEM ;
@@ -2695,10 +2702,10 @@ static int mgslpc_add_device(MGSLPC_INFO *info) current_dev->next_device = info; } -if (info->max_frame_size <4096 ) - info->max_frame_size =4096 ; -else if (info->max_frame_size >65535 ) - info->max_frame_size =65535 ; +if (info->max_frame_size <MGSLPC_MIN_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ; +else if (info->max_frame_size >MGSLPC_MAX_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MAX_FRAME_SIZE ;printk ("SyncLink PC Card %s:IO=%04X IRQ=%d\n" , info->device_name, info->io_base, info->irq_level);
Signature · 1 line
2.38.1
Quoted 9 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly. While at it, fix an edge case which could cause
rx_buf_countto be 0 whenmax_frame_sizewas set to the maximum allowed value (65535).It's worth mentioning that struct RXBUF was allocating 1 byte "too much" for what is required (ignoring bytes added by padding).
What was the result of using __packed to make sure there wasn't a sizing error?
-Kees
Quoted 112 lines
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].Signed-off-by Paulo Miguel Almeida <[email protected]>---
Changelog:- v3:
fix size calculation mistakes usingoverflow.hmacros: (Req: Andy Shevchenko, Kees Cook)
add notes struct RXBUF size (Kees Cook)- v2: removed changes to how the size of RXBUF was calculated. I changed my mind after thinking about the existing padding in the struct. Happy to discuss it if anyone sees it differently.
- v1: https://lore.kernel.org/lkml/[email protected]/
--- drivers/char/pcmcia/synclink_cs.c | 33 +++++++++++++++++++------------ 1 file changed, 20 insertions(+), 13 deletions(-)drivers/char/pcmcia/synclink_cs.c +20
-13@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */ @@ -229,12 +229,18 @@ typedef struct _mgslpc_info { }MGSLPC_INFO ;#define MGSLPC_MAGIC 0x5402 +#define MGSLPC_MAX_FRAME_SIZE 65535 +#define MGSLPC_MIN_FRAME_SIZE 4096/* * The size of the serial xmit buffer is 1 page, or 4096 bytes */ #define TXBUFSIZE 4096 +/* +* RXBUF accommodates at least 1 buffer (header+data) of MGSLPC_MAX_FRAME_SIZE +*/ +#define RXBUF_MAX_SIZE (sizeof(RXBUF) + MGSLPC_MAX_FRAME_SIZE)#define CHA 0x00/* channel A offset */ #define CHB 0x40/* channel B offset */ @@ -529,7 +535,7 @@ static int mgslpc_probe(struct pcmcia_device *link)tty_port_init (&info->port); info->port.ops = &mgslpc_port_ops;INIT_WORK (&info->task, bh_handler); - info->max_frame_size =4096 ; + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ;init_waitqueue_head (&info->status_event_wait_q);init_waitqueue_head (&info->event_wait_q);spin_lock_init (&info->lock);@@ -2611,19 +2617,20 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; +if (check_add_overflow (sizeof (RXBUF ), info->max_frame_size, &info->rx_buf_size)) +return -EINVAL ; -/* calculate total allocation size for 8 buffers */ - info->rx_buf_total_size = info->rx_buf_size *8 ; +/* try to alloc as many buffers that can fit within RXBUF_MAX_SIZE (up to 8) */ +if (check_mul_overflow (info->rx_buf_size,8 , &info->rx_buf_total_size)) +return -EINVAL ; -/* limit total allocated memory */ -if (info->rx_buf_total_size >0x10000 ) - info->rx_buf_total_size =0x10000 ; +if (info->rx_buf_total_size >RXBUF_MAX_SIZE ) + info->rx_buf_total_size =RXBUF_MAX_SIZE ;/* calculate number of buffers */ info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size; - info->rx_buf =kmalloc (info->rx_buf_total_size,GFP_KERNEL ); + info->rx_buf =kcalloc (info->rx_buf_count, info->rx_buf_size,GFP_KERNEL );if (info->rx_buf ==NULL )return -ENOMEM ;@@ -2695,10 +2702,10 @@ static int mgslpc_add_device(MGSLPC_INFO *info) current_dev->next_device = info; } -if (info->max_frame_size <4096 ) - info->max_frame_size =4096 ; -else if (info->max_frame_size >65535 ) - info->max_frame_size =65535 ; +if (info->max_frame_size <MGSLPC_MIN_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ; +else if (info->max_frame_size >MGSLPC_MAX_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MAX_FRAME_SIZE ;printk ("SyncLink PC Card %s:IO=%04X IRQ=%d\n" , info->device_name, info->io_base, info->irq_level);Signature · 1 line
2.38.1
Signature · 1 line
Kees Cook
Quoted 8 lines · Kees Cook
Quoted 7 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly. While at it, fix an edge case which could cause
rx_buf_countto be 0 whenmax_frame_sizewas set to the maximum allowed value (65535).
Quoted 8 lines
Quoted 2 lines
It's worth mentioning that struct RXBUF was allocating 1 byte "too much" for what is required (ignoring bytes added by padding).
What was the result of using
__packedto make sure there wasn't a sizing error?-Kees
With or without __packed__ attribute, sufficient space would be allocated which is good :-)
In both cases there is still some "extra space" (1 byte on __packed and 4 bytes on non-packed) but that should be negligible. OTOH, if I'm asked to cull those bytes I am happy to do it too.
pahole -C RXBUF non-packed/drivers/char/pcmcia/synclink_cs.o
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[]; /* 5 0 */
/* size: 8, cachelines: 1, members: 3 */
/* padding: 3 */
/* last cacheline: 8 bytes */
} RXBUF;pahole -C RXBUF packed/drivers/char/pcmcia/synclink_cs.o
typedef struct {
int count; /* 0 4 */
unsigned char status; /* 4 1 */
char data[]; /* 5 0 */
/* size: 5, cachelines: 1, members: 3 */
/* last cacheline: 5 bytes */
} __attribute__((__packed__)) RXBUF;- Paulo A.
Quoted 115 lines
Quoted 112 lines
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].Signed-off-by Paulo Miguel Almeida <[email protected]>---
Changelog:- v3:
fix size calculation mistakes usingoverflow.hmacros: (Req: Andy Shevchenko, Kees Cook)
add notes struct RXBUF size (Kees Cook)- v2: removed changes to how the size of RXBUF was calculated. I changed my mind after thinking about the existing padding in the struct. Happy to discuss it if anyone sees it differently.
- v1: https://lore.kernel.org/lkml/[email protected]/
--- drivers/char/pcmcia/synclink_cs.c | 33 +++++++++++++++++++------------ 1 file changed, 20 insertions(+), 13 deletions(-)drivers/char/pcmcia/synclink_cs.c +20
-13@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */ @@ -229,12 +229,18 @@ typedef struct _mgslpc_info { }MGSLPC_INFO ;#define MGSLPC_MAGIC 0x5402 +#define MGSLPC_MAX_FRAME_SIZE 65535 +#define MGSLPC_MIN_FRAME_SIZE 4096/* * The size of the serial xmit buffer is 1 page, or 4096 bytes */ #define TXBUFSIZE 4096 +/* +* RXBUF accommodates at least 1 buffer (header+data) of MGSLPC_MAX_FRAME_SIZE +*/ +#define RXBUF_MAX_SIZE (sizeof(RXBUF) + MGSLPC_MAX_FRAME_SIZE)#define CHA 0x00/* channel A offset */ #define CHB 0x40/* channel B offset */ @@ -529,7 +535,7 @@ static int mgslpc_probe(struct pcmcia_device *link)tty_port_init (&info->port); info->port.ops = &mgslpc_port_ops;INIT_WORK (&info->task, bh_handler); - info->max_frame_size =4096 ; + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ;init_waitqueue_head (&info->status_event_wait_q);init_waitqueue_head (&info->event_wait_q);spin_lock_init (&info->lock);@@ -2611,19 +2617,20 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; +if (check_add_overflow (sizeof (RXBUF ), info->max_frame_size, &info->rx_buf_size)) +return -EINVAL ; -/* calculate total allocation size for 8 buffers */ - info->rx_buf_total_size = info->rx_buf_size *8 ; +/* try to alloc as many buffers that can fit within RXBUF_MAX_SIZE (up to 8) */ +if (check_mul_overflow (info->rx_buf_size,8 , &info->rx_buf_total_size)) +return -EINVAL ; -/* limit total allocated memory */ -if (info->rx_buf_total_size >0x10000 ) - info->rx_buf_total_size =0x10000 ; +if (info->rx_buf_total_size >RXBUF_MAX_SIZE ) + info->rx_buf_total_size =RXBUF_MAX_SIZE ;/* calculate number of buffers */ info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size; - info->rx_buf =kmalloc (info->rx_buf_total_size,GFP_KERNEL ); + info->rx_buf =kcalloc (info->rx_buf_count, info->rx_buf_size,GFP_KERNEL );if (info->rx_buf ==NULL )return -ENOMEM ;@@ -2695,10 +2702,10 @@ static int mgslpc_add_device(MGSLPC_INFO *info) current_dev->next_device = info; } -if (info->max_frame_size <4096 ) - info->max_frame_size =4096 ; -else if (info->max_frame_size >65535 ) - info->max_frame_size =65535 ; +if (info->max_frame_size <MGSLPC_MIN_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ; +else if (info->max_frame_size >MGSLPC_MAX_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MAX_FRAME_SIZE ;printk ("SyncLink PC Card %s:IO=%04X IRQ=%d\n" , info->device_name, info->io_base, info->irq_level);Signature · 1 line
2.38.1Signature · 1 line
Kees Cook
Quoted 14 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly. While at it, fix an edge case which could cause
rx_buf_countto be 0 whenmax_frame_sizewas set to the maximum allowed value (65535).It's worth mentioning that struct RXBUF was allocating 1 byte "too much" for what is required (ignoring bytes added by padding).
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].
...
drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v)
Quoted 13 lines
static int
rx_alloc_buffers(MGSLPC_INFO *info){ /* each buffer has header and data */drivers/char/pcmcia/synclink_cs.c +5
-3- info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; +if (check_add_overflow (sizeof (RXBUF ), info->max_frame_size, &info->rx_buf_size)) +return -EINVAL ; -/* calculate total allocation size for 8 buffers */ - info->rx_buf_total_size = info->rx_buf_size *8 ; +/* try to alloc as many buffers that can fit within RXBUF_MAX_SIZE (up to 8) */ +if (check_mul_overflow (info->rx_buf_size,8 , &info->rx_buf_total_size)) +return -EINVAL ;
This check is implied by kcalloc(). But to make it effective we probably need to get a count first.
drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2628
Quoted 5 lines
drivers/char/pcmcia/synclink_cs.c +2
-3-/* limit total allocated memory */ -if (info->rx_buf_total_size >0x10000 ) - info->rx_buf_total_size =0x10000 ; +if (info->rx_buf_total_size >RXBUF_MAX_SIZE ) + info->rx_buf_total_size =RXBUF_MAX_SIZE ;
If max_frame_size > 8192 - sizeof(RXBUF), we bump into this condition...
Quoted 2 lines
/* calculate number of buffers */
info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size;...which means that rx_buf_count < 8...
(and if max_frame_size > RXBUF_MAX_SIZE - sizeof(RXBUF), count becomes 0, I don't know if below clamp_val() is the only place to guarantee that)
drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2633
Quoted 2 lines
drivers/char/pcmcia/synclink_cs.c +1
-1- info->rx_buf =kmalloc (info->rx_buf_total_size,GFP_KERNEL ); + info->rx_buf =kcalloc (info->rx_buf_count, info->rx_buf_size,GFP_KERNEL );
...hence rx_buf size will be less than rx_buf_total_size.
That is probably not an issue per se, but I'm wondering if the (bigger) value of rx_buf_total_size is the problem further in the code.
Quoted 2 lines
if (info->rx_buf == NULL)
return -ENOMEM;Maybe something like
static int rx_alloc_buffers(MGSLPC_INFO *info)
{
/* Prevent count from being 0 */
if (->max_frame_size > MAX_FRAME_SIZE)
return -EINVAL;
... count = ...;
...
rx_total_size = ...
rx_buf = kcalloc(...);
Then you don't need to check overflow with check_add_overflow() and check_mul_overflow() will be inside the kcalloc.
...
drivers/char/pcmcia/synclink_cs.c static int mgslpc_add_device(MGSLPC_INFO *info) · line 2708
Quoted 8 lines
drivers/char/pcmcia/synclink_cs.c +4
-4-if (info->max_frame_size <4096 ) - info->max_frame_size =4096 ; -else if (info->max_frame_size >65535 ) - info->max_frame_size =65535 ; +if (info->max_frame_size <MGSLPC_MIN_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ; +else if (info->max_frame_size >MGSLPC_MAX_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MAX_FRAME_SIZE ;
You can use clamp_val() macro here.
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 42 lines · Andy Shevchenko
Quoted 14 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly. While at it, fix an edge case which could cause
rx_buf_countto be 0 whenmax_frame_sizewas set to the maximum allowed value (65535).It's worth mentioning that struct RXBUF was allocating 1 byte "too much" for what is required (ignoring bytes added by padding).
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1]....
drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v)
Quoted 13 lines
static int
rx_alloc_buffers(MGSLPC_INFO *info){ /* each buffer has header and data */drivers/char/pcmcia/synclink_cs.c +5
-3- info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; +if (check_add_overflow (sizeof (RXBUF ), info->max_frame_size, &info->rx_buf_size)) +return -EINVAL ; -/* calculate total allocation size for 8 buffers */ - info->rx_buf_total_size = info->rx_buf_size *8 ; +/* try to alloc as many buffers that can fit within RXBUF_MAX_SIZE (up to 8) */ +if (check_mul_overflow (info->rx_buf_size,8 , &info->rx_buf_total_size)) +return -EINVAL ;This check is implied by
kcalloc(). But to make it effective we probably need to get a count first.drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2628
Quoted 5 lines
drivers/char/pcmcia/synclink_cs.c +2
-3-/* limit total allocated memory */ -if (info->rx_buf_total_size >0x10000 ) - info->rx_buf_total_size =0x10000 ; +if (info->rx_buf_total_size >RXBUF_MAX_SIZE ) + info->rx_buf_total_size =RXBUF_MAX_SIZE ;
Quoted 6 lines
If
max_frame_size> 8192 - sizeof(RXBUF), we bump into this condition...Quoted 2 lines
/* calculate number of buffers */ info->rx_buf_count = info->rx_buf_total_size / info->rx_buf_size;...which means that
rx_buf_count< 8...
that's correct. My reading of what the original author intended is the following:
- rx_buf_count can be < 8 if max_frame_size needs to be > 8192 so that userspace tools don't need to collate the different packets together then again, SyncLink_CS supports a variety of protocols.
- the more circular buffers, the better, but it looks perfectly acceptable to have 1 big rx_buf (max_frame_size possible) if the communication is orchestrated nicely (which part sends what and when) especially for RS-232-based communications.
Quoted 4 lines
(and if
max_frame_size>RXBUF_MAX_SIZE- sizeof(RXBUF), count becomes 0, I don't know if belowclamp_val()is the only place to guarantee that)
I can confirm that the clamp_val() below is the only place that guarantees the max_frame_size isn't greater than RXBUF_MAX_SIZE. That happens at the device probing stage:
( mgslpc_probe > mgslpc_add_device > clamp_val-like routine )
As max_frame_size can only be set as a module parameter and no other way is exposed to userspace to tweak that afterwards, my 2 cents is that clamp_val() routine should be fine as rx_buf_count will always be > 0 after this fix.
Quoted 9 lines
drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2633
Quoted 2 lines
drivers/char/pcmcia/synclink_cs.c +1
-1- info->rx_buf =kmalloc (info->rx_buf_total_size,GFP_KERNEL ); + info->rx_buf =kcalloc (info->rx_buf_count, info->rx_buf_size,GFP_KERNEL );...hence
rx_bufsize will be less thanrx_buf_total_size.That is probably not an issue per se, but I'm wondering if the (bigger) value of
rx_buf_total_sizeis the problem further in the code.
rx_buf_total_size isn't used outside of this function so it could be a local variable IMO.. so I would say that this wouldn't be a problem.
I had noticed that rx_buf_total_size could be moved into a local variable before but I thought that removing it from MGSLPC struct should be part of a separate patch instead.
Quoted 10 lines
Quoted 2 lines
if (info->rx_buf == NULL) return -ENOMEM;Maybe something like
static int
rx_alloc_buffers(MGSLPC_INFO *info){ /* Prevent count from being 0 */ if (->max_frame_size > MAX_FRAME_SIZE) return -EINVAL;
This boils down to whether having the clamp_val() on the probe method is sufficient in your point of view. You make the final call on this :-)
Quoted 9 lines
... count = ...; ... rx_total_size = ... rx_buf = kcalloc(...);Then you don't need to check overflow with
check_add_overflow()andcheck_mul_overflow()will be inside the kcalloc.
check_mul_overflow point -> agreed.
check_add_overflow -> similar suggestion as my previous point, if the clamp_val on probe is sufficient for you, I would say that we don't need it as of now too. But if you still think that we need it, I'm flexible with that too.
Quoted 11 lines
...
drivers/char/pcmcia/synclink_cs.c static int mgslpc_add_device(MGSLPC_INFO *info) · line 2708
Quoted 8 lines
drivers/char/pcmcia/synclink_cs.c +4
-4-if (info->max_frame_size <4096 ) - info->max_frame_size =4096 ; -else if (info->max_frame_size >65535 ) - info->max_frame_size =65535 ; +if (info->max_frame_size <MGSLPC_MIN_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MIN_FRAME_SIZE ; +else if (info->max_frame_size >MGSLPC_MAX_FRAME_SIZE ) + info->max_frame_size =MGSLPC_MAX_FRAME_SIZE ;
Quoted 2 lines
You can use
clamp_val()macro here.
Nice, I didn't know about this macro. I will make that change for v4.
All really nice points you've made Andy, I'm learning heaps of new things with this patch :-)
thanks!
- Paulo A.
Quoted 59 lines · Kees Cook
Quoted 34 lines · Paulo Miguel Almeida
One-element arrays are deprecated, and we are replacing them with flexible array members instead. So, replace one-element array with flexible-array member in struct RXBUF and refactor the rest of the code accordingly.
It's worth mentioning that doing a build before/after this patch results in no binary output differences.
This helps with the ongoing efforts to tighten the
FORTIFY_SOURCEroutines onmemcpy()and help us make progress towards globally enabling -fstrict-flex-arrays=3 [1].Signed-off-by Paulo Miguel Almeida <[email protected]>--- drivers/char/pcmcia/synclink_cs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)drivers/char/pcmcia/synclink_cs.c +1
-1@@ -105,7 +105,7 @@ static MGSL_PARAMS default_params = {typedef struct {int count;unsigned char status; -char data[1 ]; +char data[]; }RXBUF ;/* The queue of BH actions to be performed */ drivers/char/pcmcia/synclink_cs.c static int mgslpc_proc_show(struct seq_file *m, void *v) · line 2615
Quoted 6 lines
drivers/char/pcmcia/synclink_cs.c +2
-1@@ -2611,7 +2611,8 @@ static int mgslpc_proc_show(struct seq_file *m, void *v)static int rx_alloc_buffers (MGSLPC_INFO *info) {/* each buffer has header and data */ - info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size; + info->rx_buf_size =max (offsetof(typeof (RXBUF ), data) +1 ,sizeof (RXBUF )) + + info->max_frame_size;It seems like there is an existing size bug here, and likely should be fixed separately?
i.e. this was already allocating 1 byte "too much". I'd expect this first:
(quoted diff) +1
-1- info->rx_buf_size = sizeof(RXBUF) + info->max_frame_size; + info->rx_buf_size = sizeof(RXBUF) - 1 + info->max_frame_size;and then the next patch:
drivers/char/pcmcia/synclink_cs.c +2
-2-char data[1 ]; +char data[]; - info->rx_buf_size =sizeof (RXBUF ) -1 + info->max_frame_size; + info->rx_buf_size =sizeof (RXBUF ) + info->max_frame_size;
Quoted 7 lines
The above would induce a binary output change, and the second would not.
Though this results in what you had for the v2 patch (but I can't believe it had no binary changes...)
Signature · 1 line
Kees Cook
Just realised that you made a comment on PATCH v1 and Andy made a comment on PATCH v2. Please conside my answer for PATCH v2 as I have abandoned the v1. Apologies for the confusion.
thanks!
- Paulo A.