Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1175112

[PATCH 2/4] perf probe: Fix failure to probe online module

Show all headers | View raw

Kernel module has the type DSO_TYPE_USER, the check on dso type in
kernel_get_module_dso() disable openning debuginfo of online kernel
modules.

Problem can be reproduced as:

  $ insmod test_bpf.ko

  $ perf probe -v -m test_bpf --add='skb_is_nonlinear'

  probe-definition(0): skb_is_nonlinear
  symbol:skb_is_nonlinear file:(null) line:0 offset:0 return:0 lazy:(null)
  0 arguments
  Failed to find module test_bpf.
  Could not open debuginfo. Try to use symbols.
  Failed to find symbol skb_is_nonlinear in test_bpf
    Error: Failed to add events. Reason: No such file or directory (Code: -2)

Infact kernel_get_module_dso() function is called in perf-probe, in this
condition, machine->dsos list contains only dsos which is either kmod or
kernel loaded by machine__create_kernel_maps(), so there's no need to
pick away the non-kmod user type dsos.

This patch removes the check.

After this patch:

  $ perf probe -v -m test_bpf --add='skb_is_nonlinear'

  probe-definition(0): skb_is_nonlinear
  symbol:skb_is_nonlinear file:(null) line:0 offset:0 return:0 lazy:(null)
  0 arguments
  Open Debuginfo file: /lib/modules/4.1.0+/test_bpf.ko
  Try to find probe point from debuginfo.
  Matched function: skb_is_nonlinear
  found inline addr: 0x10ab3
  Probe point found: test_bpf_init+585
  Found 1 probe_trace_events.
  Opening /sys/kernel/debug/tracing/kprobe_events write=1
  Added new event:
  Writing event: p:probe/skb_is_nonlinear test_bpf:test_bpf_init+585
    probe:skb_is_nonlinear (on skb_is_nonlinear in test_bpf)

  You can now use it in all perf tools, such as:

      perf record -e probe:skb_is_nonlinear -aR sleep 1

Signed-off-by: He Kuang <hekuang@huawei.com>
---
 tools/perf/util/probe-event.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index 381f23a..8ad2743 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -275,8 +275,6 @@ static int kernel_get_module_dso(const char *module, struct dso **pdso)
 
 	if (module) {
 		list_for_each_entry(dso, &host_machine->dsos.head, node) {
-			if (!dso->kernel)
-				continue;
 			if (strncmp(dso->short_name + 1, module,
 				    dso->short_name_len - 2) == 0)
 				goto found;
-- 
1.8.5.2

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Back to linux.kernel | Previous | Next — Previous in thread | Find similar | Unroll thread

Thread

[PATCH 1/4] perf record: Use %ld for long type sample counter He Kuang <hekuang@huawei.com> - 2015-07-01 11:10 +0200
  [PATCH 4/4] perf probe: Add failure check when show variable range He Kuang <hekuang@huawei.com> - 2015-07-01 11:10 +0200
  [PATCH 3/4] perf probe: Enable --range option according to libdw version He Kuang <hekuang@huawei.com> - 2015-07-01 11:10 +0200
  [PATCH 2/4] perf probe: Fix failure to probe online module He Kuang <hekuang@huawei.com> - 2015-07-01 11:10 +0200

csiph-web