Skip to contentSkip to conversation

lib: fix bitmap_parse() on 64-bit big endian archs

PATCH RESEND In review Alexander Gordeev 2 participants 7 messages May 18, 2020 – Jun 8, 2020

Subsystems (4)
lore
Share & tools
lore

Alexander Gordeev

author maintainer
#0 lore raw

PATCH RESEND 1/1 lib: fix bitmap_parse() on 64-bit big endian archs

Commit 2d6261583be0 ("lib: rework bitmap_parse()") does not take into account order of halfwords on 64-bit big endian architectures.

Fixes 2d6261583be0 ("lib: rework bitmap_parse()")
Cc Yury Norov <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Amritha Nambiar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Chris Wilson <[email protected]>
Cc Kees Cook <[email protected]>
Cc Matthew Wilcox <[email protected]>
Cc Miklos Szeredi <[email protected]>
Cc Rasmus Villemoes <[email protected]>
Cc Steffen Klassert <[email protected]>
Cc "Tobin C . Harding" <[email protected]>
Cc Vineet Gupta <[email protected]>
Cc Will Deacon <[email protected]>
Cc Willem de Bruijn <[email protected]>
---
 lib/bitmap.c | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)
lib/bitmap.c +18 -2
index 89260aa..a725e46 100644
@@ -717,6 +717,19 @@ static const char *bitmap_get_x32_reverse(const char *start,
 	return end;
 }
 
+#if defined(__BIG_ENDIAN) && defined(CONFIG_64BIT)
+static void save_x32_chunk(unsigned long *maskp, u32 chunk, int chunk_idx)
+{
+	maskp += (chunk_idx / 2);
+	((u32 *)maskp)[(chunk_idx & 1) ^ 1] = chunk;
+}
+#else
+static void save_x32_chunk(unsigned long *maskp, u32 chunk, int chunk_idx)
+{
+	((u32 *)maskp)[chunk_idx] = chunk;
+}
+#endif
+
 /**
  * bitmap_parse - convert an ASCII hex string into a bitmap.
  * @start: pointer to buffer containing string.
@@ -738,7 +751,8 @@ int bitmap_parse(const char *start, unsigned int buflen,
 {
 	const char *end = strnchrnul(start, buflen, '\n') - 1;
 	int chunks = BITS_TO_U32(nmaskbits);
-	u32 *bitmap = (u32 *)maskp;
+	int chunk_idx = 0;
+	u32 chunk;
 	int unset_bit;
 
 	while (1) {
@@ -749,9 +763,11 @@ int bitmap_parse(const char *start, unsigned int buflen,
 		if (!chunks--)
 			return -EOVERFLOW;
 
-		end = bitmap_get_x32_reverse(start, end, bitmap++);
+		end = bitmap_get_x32_reverse(start, end, &chunk);
 		if (IS_ERR(end))
 			return PTR_ERR(end);
+
+		save_x32_chunk(maskp, chunk, chunk_idx++);
 	}
 
 	unset_bit = (BITS_TO_U32(nmaskbits) - chunks) * 32;
Signature · 1 line
1.8.3.1
#1 lore raw
Quoted 4 lines · Alexander Gordeev

Commit 2d6261583be0 ("lib: rework bitmap_parse()") does not take into account order of halfwords on 64-bit big endian architectures.

Thanks for report and the patch!

Did it work before? Can we have a test case for that that we will see the failure?

Quoted 71 lines
Fixes 2d6261583be0 ("lib: rework bitmap_parse()")
Cc Yury Norov <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Amritha Nambiar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Chris Wilson <[email protected]>
Cc Kees Cook <[email protected]>
Cc Matthew Wilcox <[email protected]>
Cc Miklos Szeredi <[email protected]>
Cc Rasmus Villemoes <[email protected]>
Cc Steffen Klassert <[email protected]>
Cc "Tobin C . Harding" <[email protected]>
Cc Vineet Gupta <[email protected]>
Cc Will Deacon <[email protected]>
Cc Willem de Bruijn <[email protected]>
---
 lib/bitmap.c | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)
lib/bitmap.c +18 -3
index 89260aa..a725e46 100644
@@ -717,6 +717,19 @@ static const char *bitmap_get_x32_reverse(const char *start,
        return end;
 }
 
+#if defined(__BIG_ENDIAN) && defined(CONFIG_64BIT)
+static void save_x32_chunk(unsigned long *maskp, u32 chunk, int chunk_idx)
+{
+       maskp += (chunk_idx / 2);
+       ((u32 *)maskp)[(chunk_idx & 1) ^ 1] = chunk;
+}
+#else
+static void save_x32_chunk(unsigned long *maskp, u32 chunk, int chunk_idx)
+{
+       ((u32 *)maskp)[chunk_idx] = chunk;
+}
+#endif
+
 /**
  * bitmap_parse - convert an ASCII hex string into a bitmap.
  * @start: pointer to buffer containing string.
@@ -738,7 +751,8 @@ int bitmap_parse(const char *start, unsigned int buflen,
 {
        const char *end = strnchrnul(start, buflen, '\n') - 1;
        int chunks = BITS_TO_U32(nmaskbits);
-       u32 *bitmap = (u32 *)maskp;
+       int chunk_idx = 0;
+       u32 chunk;
        int unset_bit;
 
        while (1) {
@@ -749,9 +763,11 @@ int bitmap_parse(const char *start, unsigned int buflen,
                if (!chunks--)
                        return -EOVERFLOW;
 
-               end = bitmap_get_x32_reverse(start, end, bitmap++);
+               end = bitmap_get_x32_reverse(start, end, &chunk);
                if (IS_ERR(end))
                        return PTR_ERR(end);
+
+               save_x32_chunk(maskp, chunk, chunk_idx++);
        }
 
        unset_bit = (BITS_TO_U32(nmaskbits) - chunks) * 32;
--

1.8.3.1

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#2 lore raw
Quoted 4 lines · Andy Shevchenko
Quoted 1 line · Alexander Gordeev
Quoted 8 lines
Quoted 3 lines

Commit 2d6261583be0 ("lib: rework bitmap_parse()") does not take into account order of halfwords on 64-bit big endian architectures.

Thanks for report and the patch!

Did it work before? Can we have a test case for that that we will see the failure?

Sorry, I wasn't clear enough. I meant a test case that is arch-independent for this very issue (so, that I can run it on LE 64 machine and see a problem).

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author maintainer
#3 lore raw
In reply to Andy Shevchenko
Quoted 11 lines · Andy Shevchenko
Quoted 4 lines · Alexander Gordeev

Commit 2d6261583be0 ("lib: rework bitmap_parse()") does not take into account order of halfwords on 64-bit big endian architectures.

Thanks for report and the patch!

Did it work before? Can we have a test case for that that we will see the failure?

The test exists and enabled with CONFIG_TEST_BITMAP. It does not appear ever passed before on 64 BE. It does not fail on 64 LE for me either.

Thanks!

[...]

Quoted 3 lines
Signature · 2 lines
With Best Regards,
Andy Shevchenko
author maintainer
#4 lore raw
Quoted 8 lines · Alexander Gordeev
Quoted 7 lines · Andy Shevchenko
Quoted 4 lines · Alexander Gordeev

Commit 2d6261583be0 ("lib: rework bitmap_parse()") does not take into account order of halfwords on 64-bit big endian architectures.

Quoted 8 lines
Quoted 4 lines

Thanks for report and the patch!

Did it work before? Can we have a test case for that that we will see the failure?

The test exists and enabled with CONFIG_TEST_BITMAP. It does not appear ever passed before on 64 BE. It does not fail on 64 LE for me either.

Hi Andy et al,

Any feedback on the fix? Does it work on 64 LE for you?

Thanks!

Quoted 7 lines

Thanks!

[...]

Quoted 3 lines
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#5 lore raw
Quoted 14 lines · Alexander Gordeev
Quoted 13 lines · Alexander Gordeev
Quoted 11 lines · Andy Shevchenko
Quoted 4 lines · Alexander Gordeev

Commit 2d6261583be0 ("lib: rework bitmap_parse()") does not take into account order of halfwords on 64-bit big endian architectures.

Thanks for report and the patch!

Did it work before? Can we have a test case for that that we will see the failure?

Quoted 7 lines
Quoted 3 lines

The test exists and enabled with CONFIG_TEST_BITMAP. It does not appear ever passed before on 64 BE. It does not fail on 64 LE for me either.

Hi Andy et al,

Any feedback on the fix? Does it work on 64 LE for you?

Test case, please.

Yes, you can simulate BE test case on LE platform and vise versa.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author maintainer
#6 lore raw
Quoted 1 line · Andy Shevchenko

Test case, please.

Hi Andy,

Below is the output of the runtime kernel test "test_bitmap". I resent the patch with Andrew Morton on CC, but did not include the excessive test output:

test_bitmap: parse: 4: input is 1, result is 0x100000000, expected 0x1
test_bitmap: parse: 5: input is deadbeef, result is 0xdeadbeef00000000, expected 0xdeadbeef
test_bitmap: parse: 6: input is 1,0, result is 0x1, expected 0x100000000
test_bitmap: parse: 7: input is deadbeef,,0,1, result is 0x1, expected 0xdeadbeef
test_bitmap: parse: 8: input is deadbeef,1,0, result is 0x1, expected 0x100000000
test_bitmap: parse: 9: input is baadf00d,deadbeef,1,0, result is 0x1, expected 0x100000000
test_bitmap: parse: 10: input is badf00d,deadbeef,1,0, errno is -75, expected 0
test_bitmap: parse: 11: input is badf00d,deadbeef,1,0, errno is -75, expected 0
test_bitmap: parse: 12: input is badf00d,deadbeef,1,0 , errno is -75, expected 0
test_bitmap: parse: 13: input is , badf00d,deadbeef,1,0 , , errno is -75, expected 0
test_bitmap: parse: 14: input is , badf00d, ,, ,,deadbeef,1,0 , , errno is -75, expected 0
test_bitmap: parse: 16: input is 3,0, errno is 0, expected -75
test_bitmap: parse_user: 4: input is 1, result is 0x100000000, expected 0x1
test_bitmap: parse_user: 5: input is deadbeef, result is 0xdeadbeef00000000, expected 0xdeadbeef
test_bitmap: parse_user: 6: input is 1,0, result is 0x1, expected 0x100000000
test_bitmap: parse_user: 7: input is deadbeef,,0,1, result is 0x1, expected 0xdeadbeef
test_bitmap: parse_user: 8: input is deadbeef,1,0, result is 0x1, expected 0x100000000
test_bitmap: parse_user: 9: input is baadf00d,deadbeef,1,0, result is 0x1, expected 0x100000000
test_bitmap: parse_user: 10: input is badf00d,deadbeef,1,0, errno is -75, expected 0
test_bitmap: parse_user: 11: input is badf00d,deadbeef,1,0, errno is -75, expected 0
test_bitmap: parse_user: 12: input is badf00d,deadbeef,1,0 , errno is -75, expected 0
test_bitmap: parse_user: 13: input is , badf00d,deadbeef,1,0 , , errno is -75, expected 0
test_bitmap: parse_user: 14: input is , badf00d, ,, ,,deadbeef,1,0 , , errno is -75, expected 0
test_bitmap: parse_user: 16: input is 3,0, errno is 0, expected -75

Thanks!

Quoted 7 lines

Yes, you can simulate BE test case on LE platform and vise versa.

Signature · 2 lines
With Best Regards,
Andy Shevchenko