perf callchain: Stop validating callchains by the max_stack sysctl
authorArnaldo Carvalho de Melo <acme@redhat.com>
Thu, 19 May 2016 14:14:15 +0000 (11:14 -0300)
committerArnaldo Carvalho de Melo <acme@redhat.com>
Fri, 20 May 2016 14:43:56 +0000 (11:43 -0300)
As thread__resolve_callchain_sample can be used for handling perf.data
files, that could've been recorded with a large max_stack sysctl setting
than what the system used for analysis has set.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Milian Wolff <milian.wolff@kdab.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Zefan Li <lizefan@huawei.com>
Link: http://lkml.kernel.org/n/tip-2995bt2g5yq2m05vga4kip6m@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools/perf/util/machine.c

index bdc33ce40bdcd93e86d3ff70e6016c6089151538..205d2701736179a40868ecb9d285a4a48d007b3b 100644 (file)
@@ -1771,11 +1771,6 @@ static int resolve_lbr_callchain_sample(struct thread *thread,
                 */
                int mix_chain_nr = i + 1 + lbr_nr + 1;
 
-               if (mix_chain_nr > (int)sysctl_perf_event_max_stack + PERF_MAX_BRANCH_DEPTH) {
-                       pr_warning("corrupted callchain. skipping...\n");
-                       return 0;
-               }
-
                for (j = 0; j < mix_chain_nr; j++) {
                        if (callchain_param.order == ORDER_CALLEE) {
                                if (j < i + 1)
@@ -1815,7 +1810,7 @@ static int thread__resolve_callchain_sample(struct thread *thread,
        struct ip_callchain *chain = sample->callchain;
        int chain_nr = chain->nr;
        u8 cpumode = PERF_RECORD_MISC_USER;
-       int i, j, err, nr_entries, nr_contexts;
+       int i, j, err, nr_entries;
        int skip_idx = -1;
        int first_call = 0;
 
@@ -1830,8 +1825,7 @@ static int thread__resolve_callchain_sample(struct thread *thread,
         * Based on DWARF debug information, some architectures skip
         * a callchain entry saved by the kernel.
         */
-       if (chain_nr < sysctl_perf_event_max_stack)
-               skip_idx = arch_skip_callchain_idx(thread, chain);
+       skip_idx = arch_skip_callchain_idx(thread, chain);
 
        /*
         * Add branches to call stack for easier browsing. This gives
@@ -1891,7 +1885,7 @@ static int thread__resolve_callchain_sample(struct thread *thread,
        }
 
 check_calls:
-       for (i = first_call, nr_entries = 0, nr_contexts = 0;
+       for (i = first_call, nr_entries = 0;
             i < chain_nr && nr_entries < max_stack; i++) {
                u64 ip;
 
@@ -1906,13 +1900,8 @@ check_calls:
 #endif
                ip = chain->ips[j];
 
-               if (ip >= PERF_CONTEXT_MAX) {
-                       if (++nr_contexts > sysctl_perf_event_max_contexts_per_stack)
-                               goto out_corrupted_callchain;
-               } else {
-                       if (++nr_entries > sysctl_perf_event_max_stack)
-                               goto out_corrupted_callchain;
-               }
+               if (ip < PERF_CONTEXT_MAX)
+                       ++nr_entries;
 
                err = add_callchain_ip(thread, cursor, parent, root_al, &cpumode, ip);
 
@@ -1921,10 +1910,6 @@ check_calls:
        }
 
        return 0;
-
-out_corrupted_callchain:
-       pr_warning("corrupted callchain. skipping...\n");
-       return 0;
 }
 
 static int unwind_entry(struct unwind_entry *entry, void *arg)