Re: [PATCH v4] tools/perf: Fix the mask in regs_dump__printf and print_sample_iregs

From: Yury Norov
Date: Thu Jun 23 2016 - 01:18:42 EST


On Thu, Jun 23, 2016 at 10:31:16AM +0530, Madhavan Srinivasan wrote:
> When decoding the perf_regs mask in regs_dump__printf(),
> we loop through the mask using find_first_bit and find_next_bit functions.
> "mask" is of type "u64", but sent as a "unsigned long *" to
> lib functions along with sizeof().
>
> While the exisitng code works fine in most of the case,
> the logic is broken when using a 32bit perf on a 64bit kernel (Big Endian).
> When reading u64 using (u32 *)(&val)[0], perf (lib/find_*_bit()) assumes it gets
> lower 32bits of u64 which is wrong. Proposed fix is to swap the words
> of the u64 to handle this case. This is _not_ endianess swap.
>
> Suggested-by: Yury Norov <ynorov@xxxxxxxxxxxxxxxxxx>
> Cc: Yury Norov <ynorov@xxxxxxxxxxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> Cc: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> Cc: Adrian Hunter <adrian.hunter@xxxxxxxxx>
> Cc: Kan Liang <kan.liang@xxxxxxxxx>
> Cc: Wang Nan <wangnan0@xxxxxxxxxx>
> Cc: Michael Ellerman <mpe@xxxxxxxxxxxxxx>
> Signed-off-by: Madhavan Srinivasan <maddy@xxxxxxxxxxxxxxxxxx>
> ---
> Changelog v3:
> 1)Moved the swap function to lib/bitmap.c
> 2)Added a macro for declaration
> 3)Added the comments
>
> Changelog v2:
> 1)Moved the swap code to a common function
> 2)Added more comments in the code
>
> Changelog v1:
> 1)updated commit message and patch subject
> 2)Add the fix to print_sample_iregs() in builtin-script.c
>
> tools/include/linux/bitmap.h | 5 +++++
> tools/lib/bitmap.c | 18 ++++++++++++++++++
> tools/perf/builtin-script.c | 4 +++-
> tools/perf/util/session.c | 4 +++-
> 4 files changed, 29 insertions(+), 2 deletions(-)
>
> diff --git a/tools/include/linux/bitmap.h b/tools/include/linux/bitmap.h
> index 28f5493da491..6be9a7ddcb03 100644
> --- a/tools/include/linux/bitmap.h
> +++ b/tools/include/linux/bitmap.h
> @@ -2,14 +2,19 @@
> #define _PERF_BITOPS_H
>
> #include <string.h>
> +#include <limits.h>
> #include <linux/bitops.h>
>
> #define DECLARE_BITMAP(name,bits) \
> unsigned long name[BITS_TO_LONGS(bits)]
>
> +#define DECLARE_U64_BITMAP(__name) \
> + unsigned long __name[sizeof(u64)/sizeof(unsigned long)]
> +
> int __bitmap_weight(const unsigned long *bitmap, int bits);
> void __bitmap_or(unsigned long *dst, const unsigned long *bitmap1,
> const unsigned long *bitmap2, int bits);
> +void bitmap_from_u64(unsigned long *dst, u64 mask);
>
> #define BITMAP_FIRST_WORD_MASK(start) (~0UL << ((start) & (BITS_PER_LONG - 1)))
>
> diff --git a/tools/lib/bitmap.c b/tools/lib/bitmap.c
> index 0a1adc1111fd..464a0cc63e6a 100644
> --- a/tools/lib/bitmap.c
> +++ b/tools/lib/bitmap.c
> @@ -29,3 +29,21 @@ void __bitmap_or(unsigned long *dst, const unsigned long *bitmap1,
> for (k = 0; k < nr; k++)
> dst[k] = bitmap1[k] | bitmap2[k];
> }
> +
> +/*
> + * bitmap_from_u64 - Check and swap words within u64.
> + * @mask: source bitmap
> + * @dst: destination bitmap
> + *
> + * In 32 bit big endian userspace on a 64bit kernel, 'unsigned long' is 32 bits.
> + * When reading u64 using (u32 *)(&val)[0] and (u32 *)(&val)[1],
> + * we will get wrong value for the mask. That is "(u32 *)(&val)[0]"
> + * gets upper 32 bits of u64, but perf may expect lower 32bits of u64.
> + */
> +void bitmap_from_u64(unsigned long *dst, u64 mask)
> +{
> + dst[0] = mask & ULONG_MAX;
> +
> + if (sizeof(mask) > sizeof(unsigned long))
> + dst[1] = mask >> 32;
> +}
> diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
> index e3ce2f34d3ad..1120ca117071 100644
> --- a/tools/perf/builtin-script.c
> +++ b/tools/perf/builtin-script.c
> @@ -412,11 +412,13 @@ static void print_sample_iregs(struct perf_sample *sample,
> struct regs_dump *regs = &sample->intr_regs;
> uint64_t mask = attr->sample_regs_intr;
> unsigned i = 0, r;
> + DECLARE_U64_BITMAP(_mask);

I thought again, and realized that it may be just
DECLARE_BITMAP(_mask, 64);

I think it's better than introduce new macro and I'd recommend you to
send v5 doing this. But this version is OK to me as well. So it's up
to you.

Reviewed-by: Yury Norov <ynorov@xxxxxxxxxxxxxxxxxx>

> if (!regs)
> return;
>
> - for_each_set_bit(r, (unsigned long *) &mask, sizeof(mask) * 8) {
> + bitmap_from_u64(_mask, mask);
> + for_each_set_bit(r, _mask, sizeof(mask) * 8) {
> u64 val = regs->regs[i++];
> printf("%5s:0x%"PRIx64" ", perf_reg_name(r), val);
> }
> diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
> index 5214974e841a..fab1f9c1e0f5 100644
> --- a/tools/perf/util/session.c
> +++ b/tools/perf/util/session.c
> @@ -940,8 +940,10 @@ static void branch_stack__printf(struct perf_sample *sample)
> static void regs_dump__printf(u64 mask, u64 *regs)
> {
> unsigned rid, i = 0;
> + DECLARE_U64_BITMAP(_mask);
>
> - for_each_set_bit(rid, (unsigned long *) &mask, sizeof(mask) * 8) {
> + bitmap_from_u64(_mask, mask);
> + for_each_set_bit(rid, _mask, sizeof(mask) * 8) {
> u64 val = regs[i++];
>
> printf(".... %-5s 0x%" PRIx64 "\n",
> --
> 1.9.1