From: Jan Stancek <jstancek@redhat.com>
To: ltp@lists.linux.it
Subject: [LTP] [PATCH v2] mtest06/mmap1: rewrite to newlib
Date: Fri, 30 Nov 2018 04:46:32 -0500 (EST) [thread overview]
Message-ID: <2002461812.80702248.1543571192512.JavaMail.zimbra@redhat.com> (raw)
In-Reply-To: <20181129134411.GA22216@rei.lan>
----- Original Message -----
> Hi!
> > Instead each mmap/munmap increases a map/unmap counter. Upon hitting
> > SIGSEGV or when comparing read value, these counter values are used
> > to determine state of mapped area as observed by first thread.
> > This isn't 100% accurrate as first thread might be faster than the
> > check, but it allows second thread to race against map/unmap for
> > its entire duration.
>
> Looks good to me, using atomic counters and comparing values before and
> after we access the memory is very clever as well. You can add my
> Reviewed-by.
>
> Very minor comments below.
> > +/* compare "before read" counters with "after read" counters */
> > +static inline int was_area_mapped(int br_m, int br_u, int ar_m, int ar_u)
> > +{
> > + return (br_m == ar_m && br_u == ar_u && br_m > br_u);
> > +}
>
> Since the br_map and br_unmap are global I would consider passing only
> the values after to this function.
Hi,
I find it more clear if it doesn't depend on globals. Since it's inlined
there should be no penalty, so I kept it as is.
> > +
> > +void *read_mem(LTP_ATTRIBUTE_UNUSED void *ptr)
> > +{
> > + int i, j, ar_map, ar_unmap;
> > + unsigned char c;
> > +
> > + for (i = 0; i < num_iter; i++) {
> > + if (setjmp(jmpbuf) == 1)
> > + continue;
I made 'i' volatile, since longjmp could clobber it.
> > +static void run(void)
> > +{
> > + pthread_t thid[2];
> > + long chld_args[1];
> > + int remaining = tst_timeout_remaining();
> > + int elapsed = 0;
> > +
> > + while (tst_timeout_remaining() > STOP_THRESHOLD) {
> > + int fd = mkfile(file_size);
> > +
> > + tst_atomic_store(0, &mapcnt);
> > + tst_atomic_store(0, &unmapcnt);
> > +
> > + chld_args[0] = fd;
>
> I would have just casted the fd here to long or intptr_t (to make sure
> compiler padds it with zeroes) then to (void*) and passed the value
> directly, i.e. (void*)(inptr_t)fd and then back in the thread we do
> fd = (intptr_t)ptr.
I changed this to pass int ptr directly, which requires no casts AFAICT.
... and pushed.
Thanks,
Jan
next prev parent reply other threads:[~2018-11-30 9:46 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-26 8:21 [LTP] [PATCH v2] mtest06/mmap1: rewrite to newlib Jan Stancek
2018-11-27 6:37 ` Li Wang
2018-11-27 8:54 ` Jan Stancek
2018-11-29 13:44 ` Cyril Hrubis
2018-11-30 9:46 ` Jan Stancek [this message]
2019-02-04 8:34 ` Naresh Kamboju
2019-02-04 9:05 ` Jan Stancek
2019-02-05 6:46 ` Jan Stancek
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=2002461812.80702248.1543571192512.JavaMail.zimbra@redhat.com \
--to=jstancek@redhat.com \
--cc=ltp@lists.linux.it \
/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