tracing: silence GCC 9 array bounds warning
PATCH 1/1 tracing: silence GCC 9 array bounds warning
Starting with GCC 9, -Warray-bounds detects cases when memset is called starting on a member of a struct but the size to be cleared ends up writing over further members.
Such a call happens in the trace code to clear, at once, all members after and including seq on struct trace_iterator:
In function 'memset',
inlined from 'ftrace_dump' at kernel/trace/trace.c:8914:3:
./include/linux/string.h:344:9: warning: '__builtin_memset' offset [8505, 8560] from the object at 'iter' is out of the bounds of referenced subobject 'seq' with type 'struct trace_seq' at offset 4368 [-Warray-bounds]
344 | return __builtin_memset(p, c, size);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
In order to avoid GCC complaining about it, we compute the address ourselves by adding the offsetof distance instead of referring directly to the member.
Since there are two places doing this clear (trace.c and trace_kdb.c), take the chance to move the workaround into a single place in the internal header.
--- kernel/trace/trace.c | 6 +----- kernel/trace/trace.h | 18 ++++++++++++++++++ kernel/trace/trace_kdb.c | 6 +----- 3 files changed, 20 insertions(+), 10 deletions(-)
kernel/trace/trace.c +1 -5
@@ -8910,12 +8910,8 @@ void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) cnt++; -/* reset all but tr, trace, and overruns */ -memset (&iter.seq,0 , -sizeof (struct trace_iterator ) - - offsetof(struct trace_iterator , seq)); +trace_iterator_reset (&iter); iter.iter_flags |=TRACE_FILE_LAT_FMT ; - iter.pos =-1 ;if (trace_find_next_entry_inc (&iter) !=NULL ) {int ret;
kernel/trace/trace.h +18 -0
@@ -1967,4 +1967,22 @@ static inline void tracer_hardirqs_off(unsigned long a0, unsigned long a1) { }extern struct trace_iterator *tracepoint_print_iter; +/* +* Reset the state of the trace_iterator so that it can read consumed data. +* Normally, the trace_iterator is used for reading the data when it is not +* consumed, and must retain state. +*/ +static __always_inline void trace_iterator_reset(struct trace_iterator *iter) +{ +const size_t offset = offsetof(struct trace_iterator , seq); + +/* +* Keep gcc from complaining about overwriting more than just one +* member in the structure. +*/ +memset ((char *)(iter) + offset,0 ,sizeof (struct trace_iterator ) - offset); + + iter->pos =-1 ; +} +#endif /* _LINUX_KERNEL_TRACE_H */
kernel/trace/trace_kdb.c +1 -5
@@ -37,12 +37,8 @@ static void ftrace_dump_buf(int skip_entries, long cpu_file)if (skip_entries)kdb_printf ("(skipping %d entries)\n" , skip_entries); -/* reset all but tr, trace, and overruns */ -memset (&iter.seq,0 , -sizeof (struct trace_iterator ) - - offsetof(struct trace_iterator , seq)); +trace_iterator_reset (&iter); iter.iter_flags |=TRACE_FILE_LAT_FMT ; - iter.pos =-1 ;if (cpu_file ==RING_BUFFER_ALL_CPUS ) {for_each_tracing_cpu (cpu) {
Signature · 1 line
2.17.1
Steven Rostedt
maintainerQuoted 4 lines · On Thu, 23 May 2019 14:45:35 +0200 Miguel Ojeda…
kernel/trace/trace.h +4
-0+/* +* Reset the state of the trace_iterator so that it can read consumed data. +* Normally, the trace_iterator is used for reading the data when it is not +* consumed, and must retain state.
kernel/trace/trace.h static inline void tracer_hardirqs_off(unsigned long a0, unsigned long a1) { } · line 1983
Quoted 10 lines
kernel/trace/trace.h +10
-0+*/ +static __always_inline void trace_iterator_reset(struct trace_iterator *iter) +{ +const size_t offset = offsetof(struct trace_iterator , seq); + +/* +* Keep gcc from complaining about overwriting more than just one +* member in the structure. +*/ +memset ((char *)(iter) + offset,0 ,sizeof (struct trace_iterator ) - offset);
I still prefer the typecast of void *, as that's used a bit more in the kernel, but since char * is also used (not as much), I'll leave it. But the parenthesis around iter are unnecessary. I'll remove them.
-- Steve
Quoted 4 lines
kernel/trace/trace.h +4
-0+ + iter->pos =-1 ; +} +
Quoted 7 lines · Steven Rostedt
On Thu, 23 May 2019 14:45:35 +0200
Miguel Ojeda <[email protected]> wrote:I still prefer the typecast of void *, as that's used a bit more in the kernel, but since char * is also used (not as much), I'll leave it. But the parenthesis around iter are unnecessary. I'll remove them.
If the preferred style in the kernel is void *, change it on your side, please! :-) Maybe we should mention it in the coding guidelines.
Cheers,
Miguel
Steven Rostedt
maintainerQuoted 5 lines · On Fri, 24 May 2019 06:05:36 +0200 Miguel Ojeda…
Quoted 4 lines · Steven Rostedt
On Thu, 23 May 2019 14:45:35 +0200
Miguel Ojeda <[email protected]> wrote:
Quoted 7 lines
Quoted 3 lines
I still prefer the typecast of void *, as that's used a bit more in the kernel, but since char * is also used (not as much), I'll leave it. But the parenthesis around iter are unnecessary. I'll remove them.
If the preferred style in the kernel is void *, change it on your side, please! :-) Maybe we should mention it in the coding guidelines.
Well, it's not officially the preferred style. There's 240 uses of (void *) in memset, compared to 98 uses of (char *)
$ git grep memset | grep '(void \*)' | wc -l 240
$ git grep memset | grep '(char \*)' | wc -l 98
I was about to make the conversion, but when I added addition to the equation, the (char *) went ahead slightly:
$ git grep memset | grep '(void \*)' | grep '+' | wc -l 32
$ git grep memset | grep '(char \*)' | grep '+' | wc -l 35
Both are fine and legit, but as the weight is slightly higher doing byte arithmetic with char *, I'll keep it as is.
-- STeve