Hi Arnaldo,
On 06/20/2018 07:19 PM, Arnaldo Carvalho de Melo wrote:
> Em Wed, Jun 20, 2018 at 07:00:30PM +0530, Ravi Bangoria escreveu:
>> perf_event__process_feature() tries to access feat_ops[feat].process
>> which is not defined for feat = HEADER_LAST_FEATURE and thus perf is
&
Hi Arnaldo,
On 06/20/2018 07:22 PM, Arnaldo Carvalho de Melo wrote:
> Em Wed, Jun 20, 2018 at 07:00:29PM +0530, Ravi Bangoria escreveu:
>> perf script in pipped mode is crashing because evsel->priv is not
>> set properly. Fix it.
>>
>> Before:
>> # ./pe
ls 2282 1031.73: 25 cpu-clock:uhH: 7effe4b3a650
Signed-off-by: Ravi Bangoria
Fixes: a14390fde64e ("perf script: Allow creating per-event dump files")
---
tools/perf/builtin-script.c | 14 ++
1 file changed, 14 insertions(+)
diff --git a/tools/perf/builtin-script.c
record -o - ls | ./perf script
Segmentation fault (core dumped)
After:
# ./perf record -o - ls | ./perf script
ls 7031 4392.099856: 25 cpu-clock:uhH: 7f5e0ce7cd60
ls 7031 4392.100355: 25 cpu-clock:uhH: 7f5e0c706ef7
Signed-off-by: Ravi Bangoria
Fixes: 57b5de463925 ("perf r
Few fields are missing in a perf script -F hint. Add them.
Signed-off-by: Ravi Bangoria
---
tools/perf/builtin-script.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index a31d7082188e..f3fefbcc4503 100644
First patch fixes perf output field hint. Second and third fixes
crash when used in a piped mode.
v2 changes:
- [PATCH 3/3] Changed as suggested by Arnaldo
v1: https://lkml.org/lkml/2018/6/20/538
Ravi Bangoria (3):
perf script: Add missing output fields in a hint
perf script: Fix crash
Hello Darrick,
Lockdep is reporting a deadlock with following trace. Saw this on my
powerpc vm with 4GB of ram, running Linus/master kernel. Though, I
don't have exact testcase to reproduce it. Is this something known?
[ 1797.620389] ==
[
Thanks Oleg for the review,
On 05/24/2018 09:56 PM, Oleg Nesterov wrote:
> On 04/17, Ravi Bangoria wrote:
>>
>> @@ -941,6 +1091,9 @@ typedef bool (*filter_func_t)(struct uprobe_consumer
>> *self,
>> if (ret)
>> goto err_buffer;
&g
Hi Srikar,
On 07/02/2018 08:24 PM, Srikar Dronamraju wrote:
> * Ravi Bangoria [2018-06-28 10:52:08]:
>
>> Reference counter gate the invocation of probe. If present,
>> by default reference count is 0. Kernel needs to increment
>> it before tracing the probe an
Hi Tvrtko,
> @@ -199,7 +199,7 @@ static inline void perf_get_data_addr(struct pt_regs
> *regs, u64 *addrp)
> if (!(mmcra & MMCRA_SAMPLE_ENABLE) || sdar_valid)
> *addrp = mfspr(SPRN_SDAR);
>
> - if (perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN) &&
> + if
Hi Oleg,
On 07/25/2018 04:38 PM, Oleg Nesterov wrote:
> No, I can't understand this patch...
>
> On 07/16, Ravi Bangoria wrote:
>>
>> --- a/kernel/events/uprobes.c
>> +++ b/kernel/events/uprobes.c
>> @@ -63,6 +63,8 @@ static struct percpu_rw_semaphore du
We assume to have only one reference counter for one uprobe.
Don't allow user to register multiple uprobes having same
inode+offset but different reference counter.
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 9 +
1 file changed, 9 insertions(+)
diff --git a/kernel
/tick
# ./perf probe sdt_tick:loop2
# ./perf stat -e sdt_tick:loop2 /tmp/tick
hi: 0
hi: 1
hi: 2
^C
Performance counter stats for '/tmp/tick':
3 sdt_tick:loop2
2.561851452 seconds time elapsed
Signed-off-by: Ravi Bangoria
Acked-by: Masami Hiramatsu
ing it as 'reference
counter' in kernel / perf code.
Signed-off-by: Ravi Bangoria
Reviewed-by: Masami Hiramatsu
[Only trace_uprobe.c]
---
include/linux/uprobes.h | 5 +
kernel/events/uprobes.c | 232 ++--
kernel/trace/trace.c| 2 +-
kernel/tr
Add addition argument 'arch_uprobe' to uprobe_write_opcode().
We need this in later set of patches.
Signed-off-by: Ravi Bangoria
---
arch/arm/probes/uprobes/core.c | 2 +-
arch/mips/kernel/uprobes.c | 2 +-
include/linux/uprobes.h| 2 +-
kernel/events/uprobes.c| 9
hi: 1
hi: 2
^C
Performance counter stats for '/tmp/tick':
3 sdt_tick:loop2
2.561851452 seconds time elapsed
Ravi Bangoria (6):
Uprobes: Simplify uprobe_register() body
Uprobe: Additional argument arch_uprobe to uprobe_write_opcode()
Uprobes:
s not conflict with
any other existing entries.
Signed-off-by: Ravi Bangoria
---
kernel/trace/trace_uprobe.c | 37 +++--
1 file changed, 35 insertions(+), 2 deletions(-)
diff --git a/kernel/trace/trace_uprobe.c b/kernel/trace/trace_uprobe.c
index bf2be098eb08..be6
Simplify uprobe_register() function body and let __uprobe_register()
handle everything. Also move dependency functions around to fix build
failures.
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 69 ++---
1 file changed, 36 insertions
Hi Thomas,
On 05/24/2018 07:26 PM, Thomas Richter wrote:
> @@ -95,7 +98,7 @@ int test__session_topology(struct test *test
> __maybe_unused, int subtest __maybe
> {
> char path[PATH_MAX];
> struct cpu_map *map;
> - int ret = -1;
> + int ret;
This is failing for me:
2600
H_SEND_CRQ 77 1762 7240 2447
Signed-off-by: Ravi Bangoria
---
.../perf/scripts/python/bin/powerpc-hcalls-record | 2 +
.../perf/scripts/python/bin/powerpc-hcalls-report | 2 +
tools/perf/scripts/python/powerpc-hcalls.py
correctness issue. I'm working on
a fix for this.
[1] https://lkml.org/lkml/2018/4/17/23
[2] https://lkml.org/lkml/2018/5/25/111
[3] https://sourceware.org/systemtap/wiki/UserSpaceProbeImplementation
[4] https://github.com/iovisor/bcc/issues/327#issuecomment-200576506
Ravi Bangoria (7
On 11/16/18 7:05 PM, Jiri Olsa wrote:
> On Fri, Nov 16, 2018 at 09:58:43AM +0530, Ravi Bangoria wrote:
>> Commit 0aa802a79469 ("perf stat: Get rid of extra clock display
>> function") introduced scale and unit for clock events. Thus,
>> perf_stat__update_shadow_
().
Reported-by: syzbot+cb1fb754b771caca0...@syzkaller.appspotmail.com
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index 96fb51f3994f..e527c4753d4f 100644
--- a/kernel/events/uprobes.c
+++ b
b/scm/linux/kernel/git/torvalds/linux.git
> Branch: master
> Target: jetson-tk1
> Lab:lab-baylibre
> Config: multi_v7_defconfig
> Plan: boot
>
> Breaking commit found:
>
> ----
...@syzkaller.appspotmail.com
Fixes: 1cc33161a83d ("uprobes: Support SDT markers having reference count
(semaphore)")
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index 32
ssues/327#issuecomment-200576506
Ravi Bangoria (4):
Uprobes: Support SDT markers having reference count (semaphore)
Uprobes/sdt: Prevent multiple reference counter for same uprobe
trace_uprobe/sdt: Prevent multiple reference counter for same uprobe
perf probe: Support SDT markers having refere
We assume to have only one reference counter for one uprobe.
Don't allow user to register multiple uprobes having same
inode+offset but different reference counter.
Signed-off-by: Ravi Bangoria
Acked-by: Srikar Dronamraju
Reviewed-by: Oleg Nesterov
---
kernel/events/uprobes.c | 19
nce counter'
in kernel / perf code.
Signed-off-by: Ravi Bangoria
Reviewed-by: Masami Hiramatsu
[Only trace_uprobe.c]
Reviewed-by: Oleg Nesterov
---
include/linux/uprobes.h | 5 +
kernel/events/uprobes.c | 259 ++--
kernel/trace/trace.c|
/tick
# ./perf probe sdt_tick:loop2
# ./perf stat -e sdt_tick:loop2 /tmp/tick
hi: 0
hi: 1
hi: 2
^C
Performance counter stats for '/tmp/tick':
3 sdt_tick:loop2
2.561851452 seconds time elapsed
Signed-off-by: Ravi Bangoria
Acked-by: Masami Hiramatsu
s not conflict with
any other existing entries.
Signed-off-by: Ravi Bangoria
Acked-by: Srikar Dronamraju
Reviewed-by: Song Liu
Reviewed-by: Oleg Nesterov
---
kernel/trace/trace_uprobe.c | 37 +++--
1 file changed, 35 insertions(+), 2 deletions(-)
diff --git a/
Hi Song,
> root@virt-test:~# ~/a.out
> 11
> semaphore 0
> semaphore 0
> semaphore 2 <<< when the uprobe is enabled
Yes, this happens when multiple vmas points to the same file portion.
Can you check /proc/`pgrep a.out`/maps.
Logic is simple. If we are going to patch an instruction,
Hi Song,
> However, if I start a.out AFTER enabling the uprobe, there is something wrong:
>
> root@virt-test:~# ~/a.out
> 11
> semaphore 0 <<< this should be non-zero, as the uprobe is already
> enabled
Ok. I'm able to reproduce this. Digging deeper.
Ravi
Hi Song,
On 08/21/2018 10:53 AM, Ravi Bangoria wrote:
> Hi Song,
>
>> However, if I start a.out AFTER enabling the uprobe, there is something
>> wrong:
>>
>> root@virt-test:~# ~/a.out
>> 11
>> semaphore 0 <<< this should be non-zero, as
On 08/21/2018 01:04 PM, Naveen N. Rao wrote:
> Song Liu wrote:
>> I am testing the patch set with the following code:
>>
>> #include
>> #include
>>
>> volatile short semaphore = 0;
>>
>> int for_uprobe(int c)
>> {
>> printf("%d\n", c + 10);
>> return c + 1;
>> }
>>
>> int
ward. Highest 40 bits of perf_event_attr.config is used to stored offset
> of the reference count (semaphore).
>
> Format information in /sys/bus/event_source/devices/uprobe/format/ is
> updated to reflect this new feature.
LGTM
Reviewed-and-tested-by: Ravi Bangoria
>
> Signed-off-
Hi Song,
On 08/11/2018 01:27 PM, Song Liu wrote:
>> +
>> +static void delayed_uprobe_delete(struct delayed_uprobe *du)
>> +{
>> + if (!du)
>> + return;
> Do we really need this check?
Not necessary though, but I would still like to keep it for a safety.
>
>> +
Hi Song,
On 08/13/2018 11:17 AM, Ravi Bangoria wrote:
>>> +
>>> +static void delayed_uprobe_remove(struct uprobe *uprobe, struct mm_struct
>>> *mm)
>>> +{
>>> + struct list_head *pos, *q;
>>> + struct de
Hi Song,
On 08/11/2018 01:42 PM, Song Liu wrote:
> Do we really need this given we already have PATCH 4/6?
> uprobe_regsiter() can be called
> out of trace_uprobe, this patch won't catch all conflicts anyway.
Right but it, at least, catch all conflicts happening via trace_uprobe.
I don't mind
Hi Oleg,
On 08/13/2018 05:20 PM, Oleg Nesterov wrote:
> On 08/13, Ravi Bangoria wrote:
>>
>> On 08/11/2018 01:27 PM, Song Liu wrote:
>>>> +
>>>> +static void delayed_uprobe_delete(struct delayed_uprobe *du)
>>>> +{
>>>> +
On 08/13/2018 06:47 PM, Oleg Nesterov wrote:
> On 08/13, Ravi Bangoria wrote:
>>
>>> But damn, process creation (exec) is trivial. We could add a new
>>> uprobe_exec()
>>> hook and avoid delayed_uprobe_install() in uprobe_mmap().
>>
>> I'm so
Hi Song,
On 08/13/2018 10:42 PM, Song Liu wrote:
> On Mon, Aug 13, 2018 at 6:17 AM, Oleg Nesterov wrote:
>> On 08/13, Ravi Bangoria wrote:
>>>
>>>> But damn, process creation (exec) is trivial. We could add a new
>>>> uprobe_exec()
>>>>
> +static int delayed_uprobe_install(struct vm_area_struct *vma)
> +{
> + struct list_head *pos, *q;
> + struct delayed_uprobe *du;
> + unsigned long vaddr;
> + int ret = 0, err = 0;
> +
> + mutex_lock(_uprobe_lock);
> + list_for_each_safe(pos, q, _uprobe_list) {
> +
447feb3 gaih_inet.constprop.7+0xf43 (/usr/lib64/libc-2.26.so)
7fff94480a13 __GI_getaddrinfo+0x163 (inlined)
13c752d6f _init+0xbfb (/usr/bin/ping)
7fff9439371f generic_start_main.isra.0+0x13f
(/usr/lib64/libc-2.26.so)
7fff9439391b __libc_start_main+0xbb (/usr/lib64/libc-2.26.so)
LGTM.
Tested-by: Ravi Bangoria
Hi Song,
On 09/25/2018 03:55 AM, Song Liu wrote:
> This patch tries to enable PMU sharing. To make perf event scheduling
> fast, we use special data structures.
>
> An array of "struct perf_event_dup" is added to the perf_event_context,
> to remember all the duplicated events under this ctx. All
Hi Thomas,
On 09/28/2018 01:13 PM, Thomas Richter wrote:
> diff --git a/tools/perf/tests/builtin-test.c b/tools/perf/tests/builtin-test.c
> index 54ca7d87236f..e83c15a95c43 100644
> --- a/tools/perf/tests/builtin-test.c
> +++ b/tools/perf/tests/builtin-test.c
> @@ -124,7 +124,7 @@ static struct
Hi Thomas,
On 09/28/2018 03:32 PM, Thomas-Mich Richter wrote:
> I can rework the patch to use the is_supported() member function. The
> down side is that the test does not show up in the list of executed tests
> anymore,
> unless some debug output is enabled:
Which should be fine because s390
ort to avoid these tests being executed on
> s390 platform:
>
> [root@s8360046 perf]# ./perf test 22
> [root@s8360046 perf]# ./perf test -v 22
> 22: Watchpoint: Disabled
> [root@s8360046 perf]#
>
> Signed-off-by: Thomas Richter
Acked-by: Ravi Bangoria
Thanks,
Ravi
Hi Milian,
Seems this has a regression:
With acme/perf/urgent:
$ ./perf record -e cycles:u --call-graph=dwarf ls
$ ./perf script
ls 13585 602082.534478: 28032 cycles:u:
1f1f4 __GI___tunables_init+0xd3dc00a4
(/usr/lib64/ld-2.26.so)
Hi Oleg,
Sorry for bit late reply.
On 08/03/2018 04:54 PM, Oleg Nesterov wrote:
> Hi Ravi,
>
> I was going to give up and ack this series, but it seems I noticed
> a bug...
>
> On 07/31, Ravi Bangoria wrote:
>>
>> +static int delayed_uprobe_add(struct uprobe
Hi Oleg,
>> I'm sorry. I didn't get this. How can uprobe go away without calling
>> uprobe_unregister()
>> -> rergister_for_each_vma()
>>-> remove_breakpoint()
>> And remove_breakpoint() will get called
>
> assuming that _unregister() will find the same vma with the probed insn.
We assume to have only one reference counter for one uprobe.
Don't allow user to register multiple uprobes having same
inode+offset but different reference counter.
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 9 +
1 file changed, 9 insertions(+)
diff --git a/kernel
ing it as 'reference
counter' in kernel / perf code.
Signed-off-by: Ravi Bangoria
Reviewed-by: Masami Hiramatsu
[Only trace_uprobe.c]
---
include/linux/uprobes.h | 5 +
kernel/events/uprobes.c | 246 ++--
kernel/trace/trace.c| 2 +-
kernel/tr
Simplify uprobe_register() function body and let __uprobe_register()
handle everything. Also move dependency functions around to fix build
failures.
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 69 ++---
1 file changed, 36 insertions
s not conflict with
any other existing entries.
Signed-off-by: Ravi Bangoria
---
kernel/trace/trace_uprobe.c | 37 +++--
1 file changed, 35 insertions(+), 2 deletions(-)
diff --git a/kernel/trace/trace_uprobe.c b/kernel/trace/trace_uprobe.c
index bf2be098eb08..be6
Add addition argument 'arch_uprobe' to uprobe_write_opcode().
We need this in later set of patches.
Signed-off-by: Ravi Bangoria
---
arch/arm/probes/uprobes/core.c | 2 +-
arch/mips/kernel/uprobes.c | 2 +-
include/linux/uprobes.h| 2 +-
kernel/events/uprobes.c| 9
/tick
# ./perf probe sdt_tick:loop2
# ./perf stat -e sdt_tick:loop2 /tmp/tick
hi: 0
hi: 1
hi: 2
^C
Performance counter stats for '/tmp/tick':
3 sdt_tick:loop2
2.561851452 seconds time elapsed
Signed-off-by: Ravi Bangoria
Acked-by: Masami Hiramatsu
sdt_tick:loop2
2.561851452 seconds time elapsed
[1] https://sourceware.org/systemtap/wiki/UserSpaceProbeImplementation
Ravi Bangoria (6):
Uprobes: Simplify uprobe_register() body
Uprobe: Additional argument arch_uprobe to uprobe_write_opcode()
Uprobes: Support SDT markers having reference
nfig_bhrb(-1) will set mmcra to -1.
Signed-off-by: Ravi Bangoria
---
arch/powerpc/perf/core-book3s.c | 6 --
arch/powerpc/perf/power8-pmu.c | 3 +++
arch/powerpc/perf/power9-pmu.c | 3 +++
3 files changed, 10 insertions(+), 2 deletions(-)
diff --git a/arch/powerpc/perf/core-book3s.c b/arc
Add a check for sample_period value sent from userspace. Negative
value does not make sense. And in powerpc arch code this could cause
a recursive PMI leading to a hang (reported when running perf-fuzzer).
Signed-off-by: Ravi Bangoria
---
kernel/events/core.c | 3 +++
1 file changed, 3
On 5/11/19 8:12 AM, Ravi Bangoria wrote:
> Consider a scenario where user creates two events:
>
> 1st event:
> attr.sample_type |= PERF_SAMPLE_BRANCH_STACK;
> attr.branch_sample_type = PERF_SAMPLE_BRANCH_ANY;
> fd = perf_event_open(attr, 0, 1, -1, 0);
>
On 5/13/19 2:26 PM, Peter Zijlstra wrote:
> On Mon, May 13, 2019 at 09:42:13AM +0200, Peter Zijlstra wrote:
>> On Sat, May 11, 2019 at 08:12:16AM +0530, Ravi Bangoria wrote:
>>> Add a check for sample_period value sent from userspace. Negative
>>> value does not make
t;
> Modify the warning message. "--user-regs=?" should be applied to show
> the available registers for --user-regs.
>
> Signed-off-by: Kan Liang
> ---
For patch 1 and 2,
Tested-by: Ravi Bangoria
Minor neat. Should we update document as well? May be something like:
t
On 3/18/19 11:56 PM, Arnaldo Carvalho de Melo wrote:
> I added this:
>
> Cc: Sukadev Bhattiprolu
> Fixes: 2a81fa3bb5ed ("perf vendor events: Add power8 PMU events")
>
> - Arnaldo
Sure. Thanks a lot Arnaldo!
perf c2c report fails if system has empty numa node(0 cpus):
$ lscpu
NUMA node0 CPU(s):
NUMA node1 CPU(s): 0-4
$ sudo ./perf c2c report
node/cpu topology bugFailed setup nodes
Fix this.
Reported-by: Nageswara R Sastry
Signed-off-by: Ravi Bangoria
---
tools/perf/util/cpumap.c
On 2/28/19 9:52 PM, Jiri Olsa wrote:
> how about attached change (untested)?
LGTM. Would you mind sending a patch.
>
> but I wonder there are some other hidden
> bugs wrt empty node
>
> jirka
>
>
> ---
> diff --git a/tools/perf/builtin-c2c.c b/tools/perf/builtin-c2c.c
> index
Thread
Thread(s) per core:4
Apart from that, for the series:
Tested-by: Ravi Bangoria
On 3/1/19 3:56 PM, Jiri Olsa wrote:
> Ravi Bangoria reported that we fail with empty
> numa node with following message:
>
> $ lscpu
> NUMA node0 CPU(s):
> NUMA node1 CPU(s): 0-4
>
> $ sudo ./perf c2c report
> node/cpu topology bugFailed setup nodes
&g
On 2/8/19 2:03 PM, Ravi Bangoria wrote:
>
>
> On 2/6/19 7:06 PM, Oleg Nesterov wrote:
>> Ravi, I am on vacation till the end of this week, can't read your patch
>> carefully.
>>
>> I am not sure I fully understand the problem, but shouldn't we change
&
On 2/7/19 3:09 PM, Mamatha Inamdar wrote:
> This patch is to remove following hardware events
> from JSON file which are not supported on POWER8
>
Acked-by: Ravi Bangoria
Hi Jiri,
On 2/1/19 1:13 PM, Jiri Olsa wrote:
> On Thu, Jan 31, 2019 at 09:27:11AM +0100, Jiri Olsa wrote:
>> On Wed, Jan 30, 2019 at 07:36:48PM +0100, Jiri Olsa wrote:
>>
>> SNIP
>>
>>> diff --git a/kernel/events/core.c b/kernel/events/core.c
>>> index 280a72b3a553..22ec63a0782e 100644
>>> ---
On 2/1/19 1:24 PM, Ravi Bangoria wrote:
> I ran fuzzer for couple of hours but I didn't see any crash with
> your previous patch.
>
> I'll try this newer one as well.
I ran fuzzer for ~8 hrs and no lockup so far.
Thanks.
On 1/21/19 12:44 AM, Tony Jones wrote:
> While updating Perf to work with Python3 and Python2 I noticed that the
> stat-cpi script was dumping core.
[...]
> Fixes: 1fcd03946b52 ("perf stat: Update per-thread shadow stats")
> Signed-off-by: Tony Jones
Tested-by: Ravi Bangoria
Hi Arnaldo,
Yes. I'm aware of it. Just that I was busy with something else so couldn't do
it.
Thanks for reminding :). Will post a patch soon.
Ravi
On 1/8/19 10:34 PM, Arnaldo Carvalho de Melo wrote:
> Hi Ravi,
>
> I noticed that in:
>
> commit
Commit aff850393200 ("powerpc: add system call table generation support")
changed how systemcall table is generated for powerpc. Incorporate these
changes into perf as well.
Signed-off-by: Ravi Bangoria
---
tools/perf/arch/powerpc/Makefile | 15 +-
.../perf/arch/pow
We use syscall.tbl to generate system call table on powerpc.
unistd.h is no longer required now. Remove it.
Signed-off-by: Ravi Bangoria
---
tools/arch/powerpc/include/uapi/asm/unistd.h | 404 ---
tools/perf/check-headers.sh | 1 -
2 files changed, 405
Hi Steve,
Please pull this patch.
Thanks.
On 11/15/18 6:13 PM, Oleg Nesterov wrote:
> On 11/15, Ravi Bangoria wrote:
>>
>> There could be a race between task exit and probe unregister:
>>
>> exit_mm()
>> mmput()
>> __mmput() upro
:
22.1: Read Only Watchpoint: Ok
22.2: Write Only Watchpoint : Ok
22.3: Read / Write Watchpoint : Ok
22.4: Modify Watchpoint : Ok
Signed-off-by: Ravi Bangoria
gt;
> On 07/16, Ravi Bangoria wrote:
>>
>> +static int delayed_uprobe_install(struct vm_area_struct *vma)
>> +{
>> +struct list_head *pos, *q;
>> +struct delayed_uprobe *du;
>> +unsigned long vaddr;
>> +int ret = 0, err = 0;
>> +
&
Oleg, Srikar, Masami, Song,
Any feedback please? :)
Thanks,
Ravi
On 07/16/2018 02:17 PM, Ravi Bangoria wrote:
> Userspace Statically Defined Tracepoints[1] are dtrace style markers
> inside userspace applications. Applications like PostgreSQL, MySQL,
> Pthread, Perl, Python, Java, Ruby
libc-2.26.so
> +8.83% _init+236 [.] _initls
> ...
>
> Reported-by: Ravi Bangoria
> Signed-off-by: Sandipan Das
Acked-by: Ravi Bangoria
Hi Song,
On 07/14/2018 05:20 AM, Song Liu wrote:
>
> Hmm... what happens when we have multiple uprobes sharing the same
> reference counter? It feels equally complicate to me. Or did I miss any
> cases here?
As far as I can think of, it should be handled by default. No special
treatment needed
Add addition argument 'arch_uprobe' to uprobe_write_opcode().
We need this in later set of patches.
Signed-off-by: Ravi Bangoria
---
arch/arm/probes/uprobes/core.c | 2 +-
arch/mips/kernel/uprobes.c | 2 +-
include/linux/uprobes.h| 2 +-
kernel/events/uprobes.c| 9
Simplify uprobe_register() function body and let __uprobe_register()
handle everything. Also move dependency functions around to fix build
failures.
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 69 ++---
1 file changed, 36 insertions
.
- No need to use FOLL_FORCE in get_user_pages_remote() while getting
a page to update reference counter. Remove it.
- Do not mention usage of reference counter in Documentation/.
v5 can be found at: https://lkml.org/lkml/2018/6/28/51
Ravi Bangoria (6):
Uprobes: Simplify uprobe_register() body
ing it as 'reference
counter' in kernel / perf code.
Signed-off-by: Ravi Bangoria
---
include/linux/uprobes.h | 5 +
kernel/events/uprobes.c | 232 ++--
kernel/trace/trace.c| 2 +-
kernel/trace/trace_uprobe.c | 38 +++-
4 files changed,
e new entry does not conflict with
any other existing entry.
2) Any one of the reference counter offset is 0 while comparing
new entry with old one.
Signed-off-by: Ravi Bangoria
---
kernel/trace/trace_uprobe.c | 40 ++--
1 file changed, 38 insertions(+), 2 dele
reference counter on his own.
Signed-off-by: Ravi Bangoria
---
kernel/events/uprobes.c | 52 +++--
1 file changed, 50 insertions(+), 2 deletions(-)
diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index 84da8512a974..563cc3e625b3 100644
/tick
# ./perf probe sdt_tick:loop2
# ./perf stat -e sdt_tick:loop2 /tmp/tick
hi: 0
hi: 1
hi: 2
^C
Performance counter stats for '/tmp/tick':
3 sdt_tick:loop2
2.561851452 seconds time elapsed
Signed-off-by: Ravi Bangoria
Acked-by: Masami Hiramatsu
Hi Paul,
On 06/06/2018 08:23 PM, Paul Clarke wrote:
> On 06/05/2018 12:50 PM, Arnaldo Carvalho de Melo wrote:
>> From: Ravi Bangoria
>>
>> Add python script to show hypervisor call statistics. Ex,
>>
>> # perf record -a -e "{powerpc:hcall_entry,powe
Hi Oleg,
On 06/08/2018 10:06 PM, Oleg Nesterov wrote:
> Hello,
>
> I am travelling till the end of the next week, can't read this version
> until I return. Just one question,
>
> On 06/06, Ravi Bangoria wrote:
>>
>> 1. One of the major reason was the deadlock
Hi Masami,
>>> Hmm, it sounds simple... maybe we can increment refctr in
>>> install_breakpoint/
>>> remove_breakpoint?
>>
>> Not really, it would be simpler if I can put it inside install_breakpoint().
>> Consider an mmap() case. Probed instruction resides in the text section
>> whereas
>>
reveu:
>>>> On Mon, Sep 10, 2018 at 03:28:11PM +0530, Ravi Bangoria wrote:
>>>>> Ex on powerpc:
>>>>> $ sudo ./perf test 22
>>>>> 22: Watchpoint:
>>>>> 22.1: Re
> While testing, I got curious, as a 'perf test' user, why one of the
> tests had the "Skip" result:
>
> [root@seventh ~]# perf test watchpoint
> 22: Watchpoint:
> 22.1: Read Only Watchpoint: Skip
> 22.2: Write Only
:
22.1: Read Only Watchpoint: Ok
22.2: Write Only Watchpoint : Ok
22.3: Read / Write Watchpoint : Ok
22.4: Modify Watchpoint : Ok
Signed-off-by: Ravi Bangoria
Hi Oleg,
On 11/14/18 9:36 PM, Oleg Nesterov wrote:
> On 11/14, Ravi Bangoria wrote:
>>
>> syzbot reported a kernel crash with delayed_uprobe_remove():
>> https://lkml.org/lkml/2018/11/1/1244
>>
>> Backtrace mentioned in the link points to a race between proc
We already have function to check if a given event is either
SW_CPU_CLOCK or SW_TASK_CLOCK. Utilize it.
Signed-off-by: Ravi Bangoria
---
tools/perf/util/stat-shadow.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat
is results in a sightly different
values of shadow stats.
Reported-by: Anton Blanchard
Fixes: 0aa802a79469 ("perf stat: Get rid of extra clock display function")
Signed-off-by: Ravi Bangoria
---
tools/perf/util/stat-shadow.c | 13 +++--
tools/perf/util/stat.h| 2 +-
2 files c
a802a79469 ("perf stat: Get rid of extra clock display function")
Signed-off-by: Ravi Bangoria
---
tools/perf/util/stat-shadow.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index f0a8cec55c47..3c22c58b3e9
Thanks Michael,
On Tuesday 22 November 2016 05:03 PM, Michael Ellerman wrote:
> Ravi Bangoria writes:
>
>> Xmon data-breakpoint feature is broken.
>>
>> Whenever there is a watchpoint match occurs, hw_breakpoint_handler will
>> be called by do_break via notifier c
10,r2,-15
▒
BTW, we have lot of such instructions in kernel.
Thanks,
-Ravi
On Monday 05 December 2016 09:26 PM, Ravi Bangoria wrote:
> Arch like powerpc has jump instructions that includes target address
> as second operand. For example, 'bne cr7,0xc00f6154'
901 - 1000 of 1366 matches
Mail list logo
