在 2016/6/8 18:15, He Kuang 写道:
The remote unwind can supported scenario where we collect on a x86_64
machine and want to do analysis on a ARM64 or x86-32 machine. Though
this is not tested, after Arnaldo questioned the above issue, I tested
and found a bug.
In util/unwind-libunwind-local.c, PE
hi, Arnaldo
On 06/15/2015 10:49 PM, Arnaldo Carvalho de Melo wrote:
Em Mon, Jun 15, 2015 at 08:06:53AM +, He Kuang escreveu:
Fix failure to probe events on arm, problem is introduced by commit
5a51fcd1f30c ("perf probe: Skip kernel symbols which is out of
.text"). For some architectures, la
ok
在 2016/5/10 0:16, Arnaldo Carvalho de Melo 写道:
Em Sat, May 07, 2016 at 08:03:46PM +0200, Jiri Olsa escreveu:
On Fri, May 06, 2016 at 08:59:07AM +, He Kuang wrote:
There's no need for dwarf support when perf recording with callchain.
Signed-off-by: He Kuang
Acked-by: Jiri Olsa
Ok, d
hi
在 2016/5/10 16:08, Adrian Hunter 写道:
On 10/05/16 10:40, He Kuang wrote:
32-bit programs can be run on 64-bit machines, so we should choose
unwind methods according to 'thread->map' instead of the host
architecture.
This patch adds methods to test whether a dso is 64-bit or 32-bit by
the cla
在 2016/5/10 18:34, Adrian Hunter 写道:
On 10/05/16 12:49, Hekuang wrote:
hi
在 2016/5/10 16:08, Adrian Hunter 写道:
On 10/05/16 10:40, He Kuang wrote:
32-bit programs can be run on 64-bit machines, so we should choose
unwind methods according to 'thread->map' instead of the host
在 2016/5/10 19:59, Adrian Hunter 写道:
On 10/05/16 14:38, Hekuang wrote:
在 2016/5/10 18:34, Adrian Hunter 写道:
On 10/05/16 12:49, Hekuang wrote:
hi
在 2016/5/10 16:08, Adrian Hunter 写道:
On 10/05/16 10:40, He Kuang wrote:
32-bit programs can be run on 64-bit machines, so we should choose
在 2016/5/10 21:44, Arnaldo Carvalho de Melo 写道:
Em Tue, May 10, 2016 at 07:40:30AM +, He Kuang escreveu:
When unwinding callchains on a different machine, vdso info should be
provided so the unwind process won't be interrupted if address falls
into vdso region.
Currently, perf does try to
在 2016/5/17 15:33, Adrian Hunter 写道:
On 13/05/16 11:51, He Kuang wrote:
There's a problem in machine__findnew_vdso(), vdso buildid generated
by a 32-bit machine stores it with the name 'vdso', but when
processing buildid on a 64-bit machine with the same 'perf.data', perf
will search for vdso
在 2016/5/16 10:50, David Ahern 写道:
On 5/15/16 7:30 PM, Hekuang wrote:
In previous patch, I use 'perf buildid-cache -a' to add vdso
binary into the HOST buildid dir.
So 'perf buildid-cache' needs the symfs option?
With this patch 'PATCH v3 3/7 UPDATE', the
在 2016/5/18 9:51, David Ahern 写道:
On 5/17/16 7:47 PM, Hekuang wrote:
在 2016/5/16 10:50, David Ahern 写道:
On 5/15/16 7:30 PM, Hekuang wrote:
In previous patch, I use 'perf buildid-cache -a' to add vdso
binary into the HOST buildid dir.
So 'perf buildid-cache' ne
hi
在 2016/5/20 0:46, Jiri Olsa 写道:
On Thu, May 19, 2016 at 11:47:38AM +, He Kuang wrote:
SNIP
#endif /* ARCH_PERF_COMMON_H */
diff --git a/tools/perf/config/Makefile b/tools/perf/config/Makefile
index 1e46277..a86b864 100644
--- a/tools/perf/config/Makefile
+++ b/tools/perf/config/Makef
在 2016/5/20 0:19, Jiri Olsa 写道:
On Thu, May 19, 2016 at 11:50:10AM -0300, Arnaldo Carvalho de Melo wrote:
Em Thu, May 19, 2016 at 11:47:38AM +, He Kuang escreveu:
Currently, perf script uses host unwind methods to parse perf.data
callchain info regardless of the target architecture. So we
ping..
在 2016/5/16 12:51, He Kuang 写道:
There's a display inconsistency when 'call-graph' config event appears
in different position. The problem can be reproduced like this:
We record signal_deliver with call-graph and signal_generate without it.
$ perf record -g -a -e signal:signal_deliver
hi
在 2016/5/27 1:42, Jiri Olsa 写道:
On Tue, May 24, 2016 at 09:20:27AM +, He Kuang wrote:
Use thread specific unwind ops to unwind cross-platform callchains.
Currently, unwind methods is suitable for local unwind, this patch
changes the fixed methods to thread/map related. Each time a map i
在 2016/5/27 15:38, Jiri Olsa 写道:
On Fri, May 27, 2016 at 03:13:04PM +0800, Hekuang wrote:
SNIP
- I understand we need to compile 3 objects from unwind-libunwind.c,
how about we create 3 files like:
util/unwind-libunwind-local.c
util/unwind-libunwind-x86_32.c
hi
在 2016/5/13 4:23, David Ahern 写道:
On 5/12/16 7:09 AM, Arnaldo Carvalho de Melo wrote:
Em Thu, May 12, 2016 at 08:43:12AM +, He Kuang escreveu:
Symfs dir and buildid dir are two places that perf looks into for
symbols, currently, if symfs dir is given, buildid-cache is skipped.
In the c
在 2016/5/12 18:06, Adrian Hunter 写道:
On 12/05/16 11:43, He Kuang wrote:
This is a preparation for cross-platform vdso lookup.
There is a naming confusion about vdso name, vdso buildid generated by
a 32-bit machine stores it with the name 'vdso', but when processing
buildid on a 64-bit machine
在 2016/5/14 22:43, David Ahern 写道:
On 5/14/16 2:19 AM, He Kuang wrote:
In the cross-platform perf record/script scenario, we need vdsos in
buildid-cache dir and other libs in symfs dir at the same time. For
the reason that to have every single file opened by perf is relative
to symfs dirctory,
在 2016/5/16 10:50, David Ahern 写道:
On 5/15/16 7:30 PM, Hekuang wrote:
In previous patch, I use 'perf buildid-cache -a' to add vdso
binary into the HOST buildid dir.
So 'perf buildid-cache' needs the symfs option?
No, for the host, we don't reference any files
Hi, jirka
On 05/12/2015 08:37 PM, Jiri Olsa wrote:
On Tue, May 12, 2015 at 06:41:56AM +, He Kuang wrote:
SNIP
$(Q)$(MAKE) $(build)=perf
-$(OUTPUT)perf: $(PERFLIBS) $(PERF_IN)
- $(QUIET_LINK)$(CC) $(CFLAGS) $(LDFLAGS) $(PERF_IN) $(LIBS) -o $@
+LD_LIBTRACEEVENT_FLAGS += -X
hi, Arnaldo
On 05/11/2015 09:58 PM, Arnaldo Carvalho de Melo wrote:
Em Mon, May 11, 2015 at 01:10:03PM +, He Kuang escreveu:
It is not easy for users to get the accurate byte offset or the line
number where a local variable can be probed. With '--range' option,
local variables in scope of t
eBPF is very flexible, which means it is bound to have someone use it
in a way you never dreamed of, and that will be what bites you in the
end (pun intended).
understood :)
let's start slow then with bpf+syscall and bpf+kprobe only.
I think BPF + system calls/kprobes can meet our use case
(
在 2015/2/12 16:07, Namhyung Kim 写道:
Hi,
On Wed, Feb 11, 2015 at 10:01:08AM +0800, He Kuang wrote:
When build with ARCH=x86_64, perf failed to compile with following error:
tests/builtin-test.o:(.data+0x158): undefined reference to
`test__perf_time_to_tsc'
collect2: error: ld returned 1 exit
hi
在 2016/3/30 18:38, Mel Gorman 写道:
On Wed, Mar 30, 2016 at 10:22:07AM +, He Kuang wrote:
This reverts commit 998d39cb236fe464af86a3492a24d2f67ee1efc2.
When local irq is disabled, a percpu variable does not change, so we can
remove the access macros and let the compiler optimize the code
Hi
在 2016/3/30 19:10, Michal Hocko 写道:
On Wed 30-03-16 18:51:12, Hekuang wrote:
hi
在 2016/3/30 18:38, Mel Gorman 写道:
On Wed, Mar 30, 2016 at 10:22:07AM +, He Kuang wrote:
This reverts commit 998d39cb236fe464af86a3492a24d2f67ee1efc2.
When local irq is disabled, a percpu variable does
hi
在 2016/3/31 9:39, Zefan Li 写道:
On 2016/3/31 9:14, Hekuang wrote:
Hi
在 2016/3/30 19:10, Michal Hocko 写道:
On Wed 30-03-16 18:51:12, Hekuang wrote:
hi
在 2016/3/30 18:38, Mel Gorman 写道:
On Wed, Mar 30, 2016 at 10:22:07AM +, He Kuang wrote:
This reverts commit
hi,
在 2016/4/1 21:13, Arnaldo Carvalho de Melo 写道:
Em Fri, Apr 01, 2016 at 03:49:32AM +, He Kuang escreveu:
Build errors on aarch64:
libperf.a(libperf-in.o): In function `convert_timestamp':
util/jitdump.c:356: undefined reference to `tsc_to_perf_time'
collect2: error: ld returned
hi
在 2017/1/25 3:09, Arnaldo Carvalho de Melo 写道:
Em Tue, Jan 24, 2017 at 06:25:18PM +, Will Deacon escreveu:
On Tue, Jan 24, 2017 at 10:30:15AM +, He Kuang wrote:
Since HAVE_KPROBES can be enabled in arm64, this patch introduces
regs_query_register_offset() to convert register name to
在 2016/6/23 10:02, Wangnan (F) 写道:
Hi,
This patch fixes a real crash problem when we do 'perf report'
on an arm64 platform with arm32 program.
It is introduced by commit f9b2bdf228 ("perf tools: Find vdso
with the consider of cross-platform"). From dmesg report, perf
crashes in dso__type() bec
hi
在 2016/6/15 21:34, Arnaldo Carvalho de Melo 写道:
Em Tue, May 17, 2016 at 09:04:54AM +, He Kuang escreveu:
There's a problem in machine__findnew_vdso(), vdso buildid generated
by a 32-bit machine stores it with the name 'vdso', but when
processing buildid on a 64-bit machine with the same
在 2016/6/16 5:29, Arnaldo Carvalho de Melo 写道:
Em Thu, Jun 16, 2016 at 12:11:04AM +0300, Yury Norov escreveu:
On Wed, Jun 15, 2016 at 10:51:27PM +0300, Yury Norov wrote:
Maybe there already is some macro doing the conversion for you...
yes it is, cpu_to_le64() is what you want
Beware that t
hi
在 2016/6/30 20:06, Arnaldo Carvalho de Melo 写道:
Hi He,
While testing a patch by Peter Zijlstra to the --stdio
annotation code I came accross these messages:
[acme@jouet linux]$ perf annotate __vdso_gettimeofday 2>&1 | head -20
unwind: target platform=x86 is not supported
unwind: tar
在 2016/6/1 16:40, Jiri Olsa 写道:
On Tue, May 31, 2016 at 11:19:11AM +, He Kuang wrote:
SNIP
diff --git a/tools/perf/util/unwind-libunwind.c
b/tools/perf/util/unwind-libunwind.c
index e183390..5774317 100644
--- a/tools/perf/util/unwind-libunwind.c
+++ b/tools/perf/util/unwind-libunwind.c
在 2016/6/1 16:40, Jiri Olsa 写道:
On Tue, May 31, 2016 at 11:19:12AM +, He Kuang wrote:
SNIP
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 7746e09..fced833 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -102,6 +102,7 @@ libperf-$(CONFIG_LIBDW_DWARF_UNW
在 2016/6/1 16:40, Jiri Olsa 写道:
On Tue, May 31, 2016 at 11:19:08AM +, He Kuang wrote:
SNIP
-int unwind__prepare_access(struct thread *thread)
+int unwind__prepare_access(struct thread *thread, struct map *map)
{
- unwind__register_ops(thread, local_unwind_libunwind_ops);
+
在 2016/5/28 20:10, Wangnan (F) 写道:
On 2016/5/28 19:59, He Kuang wrote:
This patch extracts codes related to specific arithecture out of
unwind-libunwind.c. The extrated part are only built if local
libunwind is supported.
Signed-off-by: He Kuang
---
tools/perf/util/Build
hi
在 2016/5/30 16:53, Jiri Olsa 写道:
On Sat, May 28, 2016 at 11:59:59AM +, He Kuang wrote:
Support x86(32-bit) cross platform callchain unwind.
Signed-off-by: He Kuang
---
tools/perf/arch/Build | 1 +
tools/perf/arch/x86/util/unwind-libunwind.c| 7
在 2016/5/30 17:30, Jiri Olsa 写道:
On Mon, May 30, 2016 at 05:11:35PM +0800, Hekuang wrote:
hi
在 2016/5/30 16:53, Jiri Olsa 写道:
On Sat, May 28, 2016 at 11:59:59AM +, He Kuang wrote:
Support x86(32-bit) cross platform callchain unwind.
Signed-off-by: He Kuang
---
tools/perf/arch
在 2016/6/16 19:36, Jiri Olsa 写道:
On Thu, Jun 16, 2016 at 10:41:31AM +, He Kuang wrote:
From: Wang Nan
This patch adopts the macros for byte order conversion from
"include/linux/byteorder/generic.h" to
"tools/include/linux/byteorder/generic.h"
tools/perf/MANIFEST is also updated for 'mak
hi,
在 2017/1/27 3:31, Arnaldo Carvalho de Melo 写道:
Em Thu, Jan 26, 2017 at 04:52:12PM +, Will Deacon escreveu:
On Thu, Jan 26, 2017 at 10:49:16AM +0900, Masami Hiramatsu wrote:
On Wed, 25 Jan 2017 13:32:01 +
Will Deacon wrote:
On Wed, Jan 25, 2017 at 07:23:11AM +, He Kuang wrote
hi
在 2017/2/3 21:00, Will Deacon 写道:
On Fri, Feb 03, 2017 at 11:06:05AM +, He Kuang wrote:
This patch changes the 'dwarfnum' to 'offset' in register table, so
the index of array becomes the dwarfnum (the index of each register
defined by DWARF) and the "offset" member means the byte-offset
hi
在 2016/6/8 3:44, Arnaldo Carvalho de Melo 写道:
Em Fri, Jun 03, 2016 at 09:06:29AM +0200, Jiri Olsa escreveu:
On Fri, Jun 03, 2016 at 03:33:09AM +, He Kuang wrote:
SNIP
For using remote libunwind libraries, reference this:
http://thread.gmane.org/gmane.linux.kernel/2224430
and now w
在 2016/6/8 18:15, He Kuang 写道:
The remote unwind can supported scenario where we collect on a x86_64
machine and want to do analysis on a ARM64 or x86-32 machine. Though
this is not tested, after Arnaldo questioned the above issue, I tested
and found a bug.
In util/unwind-libunwind-local.c, PE
在 2016/6/4 5:09, Arnaldo Carvalho de Melo 写道:
Em Fri, Jun 03, 2016 at 06:06:02PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Fri, Jun 03, 2016 at 04:42:05PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Fri, Jun 03, 2016 at 09:06:29AM +0200, Jiri Olsa escreveu:
On Fri, Jun 03, 2016 at 03:33:
hi
在 2016/6/27 4:48, Alexei Starovoitov 写道:
On Sun, Jun 26, 2016 at 11:20:52AM +, He Kuang wrote:
This patchset is based on Wang Nan's v1:
http://thread.gmane.org/gmane.linux.kernel/2203717/focus=2203707
"""
This patch set allows to perf invoke some user space BPF scripts on
so
hi
在 2016/6/27 5:08, Nilay Vaish 写道:
On 26 June 2016 at 06:20, He Kuang wrote:
From: Wang Nan
This patch copies "include/linux/math64.h" into
"tools/include/linux/math64.h" and copies
"include/asm-generic/div64.h" into
"tools/include/asm-generic/div64.h", to enable other libraries use
arithm
在 2016/6/27 4:48, Alexei Starovoitov 写道:
On Sun, Jun 26, 2016 at 11:20:52AM +, He Kuang wrote:
bounds check just like ubpf library does.
hmm. I don't think I suggested to hack bpf/core.c into separate file
and compile it for userspace...
Maybe I misunderstood your suggestion. Now I ju
hi
在 2016/6/28 22:57, Alexei Starovoitov 写道:
return 0;
}
@@ -465,7 +465,7 @@ EXPORT_SYMBOL_GPL(__bpf_call_base);
*
* Decode and execute eBPF instructions.
*/
-static unsigned int __bpf_prog_run(void *ctx, const struct bpf_insn *insn)
+unsigned int __bpf_prog_run(void *ctx, c
This problem can be reproduced as follows:
We know cat /proc/version will read the memory of symbol
linux_proc_banner, then we make a hardware memory access
breakpoint on that address.
on terminal 1:
$ perf record -e mem:0x$(cat /proc/kallsyms|grep
linux_proc_banner|cut -d " " -f 1):rw --no-
ping.
在 2016/8/4 19:25, He Kuang 写道:
Because perf data from pipe do not have a header with evsel attr, we
should not check that and disable symbol_conf.use_callchain. Otherwise,
perf script won't show callchains even if the data stream contains
callchain.
Before:
$ perf record -g -o - uname
50 matches
Mail list logo
