From: "Martin Liška" <mliska@suse.cz>
To: Nathan Sidwell <nathan@acm.org>, gcc-patches@gcc.gnu.org
Cc: jh@suse.cz
Subject: Re: [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch
Date: Fri, 05 Aug 2016 12:48:00 -0000 [thread overview]
Message-ID: <b62f3336-b332-048f-ed2f-9392824f0b84@suse.cz> (raw)
In-Reply-To: <67fda6d2-9b3e-a0d1-effc-34e1115030b2@acm.org>
On 08/05/2016 02:38 PM, Nathan Sidwell wrote:
> On 08/05/16 04:55, Martin Liška wrote:
>
>> Thank you for very intensive brainstorming ;) Well I still believe that following code
>> is not thread safe, let's image following situation:
>>
>
> yeah, you're right.
>
>>> we could do better by using compare_exchange storing value, and detect the race I mentioned:
>>>
>>> gcov_t expected, val;
>>> atomic_load (&counter[0], &val, ...);
>>
>> [thread 1]: value == 1, read val == 1 // scheduled here
>>
>>> gcov_t delta = val == value ? 1 : -1;
>>> atomic_add (&counter[1], delta);
>>> if (delta < 0) {
>>> retry:
>>> /* can we set counter[0]? */
>>> atomic_load (&counter[1], &expected, ...);
>>> if (expected < 0) {
>>> bool stored = atomic_compare_exchange (&counter[0], &val, &value, ...);
>>> if (!stored && val != value)
>>> goto retry;
>> [thread 2]: value == 2, just updated counter[0] to 2
>> // after that [thread 1] continue, but wrongly does counter[1]++, but value != counter[0]
>>> atomic_add (&counter[1], 2, ...);
>
> Bah. but (a) does it matter enough? and (b) if so does changing the delta<0 handling to store a count of 1 solve it?: (answer: no)
>
> gcov_t expected, val;
> A:atomic_load (&counter[0], &val, ...);
> gcov_t delta = val == value ? 1 : -1;
> B:atomic_add (&counter[1], delta);
>
> if (delta < 0) {
> /* can we set counter[0]? */
> C:atomic_load (&counter[1], &expected, ...);
> if (expected < 0) {
> D:atomic_store (&counter[0], &value);
> E: atomic_store (&counter[1], 1);
> }
> atomic_add (&counter[1], 2, ...);
>
>
> thread-1: value = 1, reads '1' at A
> thread-2: value = 2, reads '1' at A
> thread-2: decrements count @ B
> thread-2: reads -1 at C
> thread-2: write 2 at D
> thread-2: stores 1 at E
> thread-1: increments count @ B (finally)
>
> So we still can go awry. But the code's simpler. Like you said, I don't think it's possible to solve without an atomic update to both counter[0] & counter[1].
>
>
>> Well, I wrote attached test-case which should trigger a data-race, but TSAN is silent:
>
> I'm not too surprised. The race window is tiny and you put a printf in the middle of one path. I suspect if you put a sleep/printf on the counter[1] increment path you'll see it more often.
>
> nathan
Ok, after all the experimenting and inventing "almost" thread-safe code, I incline to not to include __gcov_one_value_profiler_body_atomic
counter. The final solution is cumbersome and probably does not worth doing. Moreover, even having a thread-safe implementation, result of an indirect call counter
is not going to be stable among different runs (due to a single value storage capability).
If you agree, I'll prepare a final version of patch?
Thanks,
Martin
next prev parent reply other threads:[~2016-08-05 12:48 UTC|newest]
Thread overview: 95+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-08-01 8:50 [PATCH 0/4] Various GCOV/PGO improvements marxin
2016-08-01 8:49 ` [PATCH 3/4] Fix typo in gcov.texi marxin
2016-08-01 8:50 ` [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch marxin
2016-08-01 12:22 ` Nathan Sidwell
2016-08-01 13:29 ` Martin Liška
2016-08-04 14:48 ` Nathan Sidwell
2016-08-04 15:34 ` Martin Liška
2016-08-04 16:43 ` Nathan Sidwell
2016-08-04 17:03 ` Nathan Sidwell
2016-08-05 8:55 ` Martin Liška
2016-08-05 12:38 ` Nathan Sidwell
2016-08-05 12:48 ` Martin Liška [this message]
2016-08-05 13:14 ` Nathan Sidwell
2016-08-05 13:43 ` Martin Liška
2016-08-08 13:59 ` [PATCH 5/N] Add new *_atomic counter update function, (-fprofile-update=atomic) Martin Liška
2016-08-08 15:24 ` Nathan Sidwell
2016-08-08 16:51 ` Martin Liška
2016-08-08 17:03 ` Martin Liška
2016-08-09 12:36 ` Nathan Sidwell
2016-08-08 16:56 ` [PATCH] Fix POW2 histogram Martin Liška
2016-08-09 8:41 ` [PATCH 2/N] Fix usage of " Martin Liška
2016-08-09 12:37 ` Nathan Sidwell
2016-08-09 12:34 ` [PATCH] Fix " Nathan Sidwell
2016-08-09 11:24 ` [PATCH] Set -fprofile-update=atomic when -pthread is present Martin Liška
2016-08-09 12:40 ` Nathan Sidwell
2016-08-09 19:04 ` Andi Kleen
2016-08-12 13:31 ` Martin Liška
2016-08-18 3:16 ` Jeff Law
2016-08-18 11:02 ` Nathan Sidwell
2016-08-18 15:51 ` Andi Kleen
2016-08-18 15:53 ` Jeff Law
2016-10-03 12:13 ` Martin Liška
2016-10-03 12:26 ` Nathan Sidwell
2016-10-03 16:46 ` Jeff Law
2016-10-03 17:52 ` Andi Kleen
2016-10-04 12:05 ` Martin Liška
2016-10-05 17:54 ` Jeff Law
2016-10-13 15:34 ` [PATCH] Introduce -fprofile-update=maybe-atomic Martin Liška
2016-10-31 9:13 ` Martin Liška
2016-11-10 13:19 ` Martin Liška
2016-11-10 15:43 ` Nathan Sidwell
2016-11-10 15:55 ` David Edelsohn
2016-11-10 16:18 ` Nathan Sidwell
2016-11-10 15:58 ` Martin Liška
2016-11-10 16:17 ` David Edelsohn
2016-11-10 16:24 ` Martin Liška
2016-11-10 17:31 ` Nathan Sidwell
2016-11-11 10:48 ` Martin Liška
2016-11-11 15:11 ` Nathan Sidwell
2016-11-10 16:14 ` Nathan Sidwell
2016-11-10 16:16 ` David Edelsohn
2016-08-18 15:54 ` [PATCH] Set -fprofile-update=atomic when -pthread is present Jakub Jelinek
2016-08-18 16:06 ` Richard Biener
2016-09-07 11:41 ` Martin Liška
[not found] ` <CAFiYyc0UaSzXhZmyG9QRkHGT4JFowxBfE2yb-NvXE=hR1xafdA@mail.gmail.com>
2016-09-15 10:18 ` [RFC] Speed-up -fprofile-update=atomic Martin Liška
2016-10-04 9:45 ` Richard Biener
2016-10-12 13:53 ` Martin Liška
2016-10-13 9:43 ` Richard Biener
2016-10-17 11:47 ` Martin Liška
[not found] ` <CAFiYyc3eDT4g926PPZuktz5fEW=k-PibAcxhigx4GBcxoXNJFQ@mail.gmail.com>
2016-10-24 12:09 ` Martin Liška
[not found] ` <CAFiYyc1tSdTdqqkHcMp+dgE43+8tHL6kY8E07TCHoZBeUT-ggQ@mail.gmail.com>
2016-10-25 14:32 ` Martin Liška
2016-10-26 9:29 ` Richard Biener
2016-10-26 9:32 ` Richard Biener
2016-08-18 16:04 ` [PATCH] Set -fprofile-update=atomic when -pthread is present Richard Biener
2016-08-10 12:57 ` [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch Nathan Sidwell
2016-08-13 12:14 ` [BUILDROBOT] avr broken (was: [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch) Jan-Benedict Glaw
[not found] ` <4455937b-eba7-fe66-fe1a-3172567dd1e4@suse.cz>
2016-08-16 13:36 ` [BUILDROBOT] avr broken Nathan Sidwell
[not found] ` <617e8799-b7db-fefd-b3a3-842e9a7decfd@suse.cz>
2016-08-16 14:31 ` Nathan Sidwell
2016-08-16 17:05 ` Jan-Benedict Glaw
2016-08-16 18:26 ` Nathan Sidwell
2016-08-17 7:21 ` Denis Chertykov
2016-08-17 7:22 ` Martin Liška
2016-08-17 8:11 ` Jan-Benedict Glaw
2016-08-16 12:56 ` [PATCH] Detect whether target can use -fprofile-update=atomic Martin Liška
2016-08-16 14:31 ` Nathan Sidwell
2016-09-06 10:57 ` Martin Liška
2016-09-06 11:17 ` David Edelsohn
2016-09-06 12:15 ` Nathan Sidwell
2016-09-06 12:39 ` Jakub Jelinek
2016-09-06 12:43 ` David Edelsohn
2016-09-06 12:41 ` David Edelsohn
2016-09-06 12:51 ` Martin Liška
2016-09-06 13:13 ` Jakub Jelinek
2016-09-06 13:15 ` Martin Liška
2016-09-06 13:45 ` Jakub Jelinek
2016-09-06 13:50 ` Martin Liška
2016-09-06 14:06 ` Jakub Jelinek
2016-09-07 7:52 ` Christophe Lyon
2016-09-07 9:35 ` Martin Liška
2016-09-07 16:06 ` Christophe Lyon
2016-09-12 20:20 ` Jeff Law
2016-09-29 8:31 ` Rainer Orth
2016-08-01 8:50 ` [PATCH 2/4] Remove __gcov_indirect_call_profiler marxin
2016-08-01 8:50 ` [PATCH 4/4] Add tests for __gcov_dump and __gcov_reset marxin
2016-08-01 12:11 ` [PATCH 0/4] Various GCOV/PGO improvements Nathan Sidwell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b62f3336-b332-048f-ed2f-9392824f0b84@suse.cz \
--to=mliska@suse.cz \
--cc=gcc-patches@gcc.gnu.org \
--cc=jh@suse.cz \
--cc=nathan@acm.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).