* [LTP] [PATCH] syscalls/setrlimit03.c: read /proc/sys/fs/nr_open for RLIMIT_NOFILE limit
@ 2019-01-30 15:00 Tommi Rantala
2019-01-30 16:15 ` Cyril Hrubis
0 siblings, 1 reply; 3+ messages in thread
From: Tommi Rantala @ 2019-01-30 15:00 UTC (permalink / raw
To: ltp
From: Tommi Rantala <tommi.t.rantala@nokia.com>
Since kernel v2.6.25 RLIMIT_NOFILE limit is no longer hardcoded to
NR_OPEN, but can be set via /proc/sys/fs/nr_open, see kernel commit
9cfe015aa424b3c003baba3841a60dd9b5ad319b ("get rid of NR_OPEN and
introduce a sysctl_nr_open").
nr_open default value is 1024*1024, so setrlimit03 has been passing fine
on new kernels, only "unexpectedly succeeding" if nr_open is set to some
larger value.
Signed-off-by: Tommi Rantala <tommi.t.rantala@nokia.com>
---
testcases/kernel/syscalls/setrlimit/setrlimit03.c | 14 ++++++++++++--
1 file changed, 12 insertions(+), 2 deletions(-)
diff --git a/testcases/kernel/syscalls/setrlimit/setrlimit03.c b/testcases/kernel/syscalls/setrlimit/setrlimit03.c
index 29b52aa7f..643432bb4 100644
--- a/testcases/kernel/syscalls/setrlimit/setrlimit03.c
+++ b/testcases/kernel/syscalls/setrlimit/setrlimit03.c
@@ -36,6 +36,7 @@
#endif
static struct rlimit rlim1, rlim2;
+static unsigned int nr_open;
static struct tcase {
struct rlimit *rlimt;
@@ -51,7 +52,10 @@ static void verify_setrlimit(unsigned int n)
TEST(setrlimit(RLIMIT_NOFILE, tc->rlimt));
if (TST_RET != -1) {
- tst_res(TFAIL, "call succeeded unexpectedly");
+ tst_res(TFAIL, "call succeeded unexpectedly "
+ "(nr_open=%u rlim_cur=%lu rlim_max=%lu)", nr_open,
+ (unsigned long)(tc->rlimt->rlim_cur),
+ (unsigned long)(tc->rlimt->rlim_max));
return;
}
@@ -65,10 +69,16 @@ static void verify_setrlimit(unsigned int n)
static void setup(void)
{
+ if (tst_kvercmp(2, 6, 25) < 0) {
+ nr_open = NR_OPEN;
+ } else {
+ SAFE_FILE_SCANF("/proc/sys/fs/nr_open", "%u", &nr_open);
+ }
+
SAFE_GETRLIMIT(RLIMIT_NOFILE, &rlim1);
rlim2.rlim_max = rlim1.rlim_cur;
rlim2.rlim_cur = rlim1.rlim_max + 1;
- rlim1.rlim_max = NR_OPEN + 1;
+ rlim1.rlim_max = nr_open + 1;
}
static struct tst_test test = {
--
2.19.2
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [LTP] [PATCH] syscalls/setrlimit03.c: read /proc/sys/fs/nr_open for RLIMIT_NOFILE limit
2019-01-30 15:00 [LTP] [PATCH] syscalls/setrlimit03.c: read /proc/sys/fs/nr_open for RLIMIT_NOFILE limit Tommi Rantala
@ 2019-01-30 16:15 ` Cyril Hrubis
2019-01-30 16:22 ` Tommi Rantala
0 siblings, 1 reply; 3+ messages in thread
From: Cyril Hrubis @ 2019-01-30 16:15 UTC (permalink / raw
To: ltp
Hi!
> Signed-off-by: Tommi Rantala <tommi.t.rantala@nokia.com>
> ---
> testcases/kernel/syscalls/setrlimit/setrlimit03.c | 14 ++++++++++++--
> 1 file changed, 12 insertions(+), 2 deletions(-)
>
> diff --git a/testcases/kernel/syscalls/setrlimit/setrlimit03.c b/testcases/kernel/syscalls/setrlimit/setrlimit03.c
> index 29b52aa7f..643432bb4 100644
> --- a/testcases/kernel/syscalls/setrlimit/setrlimit03.c
> +++ b/testcases/kernel/syscalls/setrlimit/setrlimit03.c
> @@ -36,6 +36,7 @@
> #endif
>
> static struct rlimit rlim1, rlim2;
> +static unsigned int nr_open;
I would have initialized the variable to NR_OPEN here.
> static struct tcase {
> struct rlimit *rlimt;
> @@ -51,7 +52,10 @@ static void verify_setrlimit(unsigned int n)
>
> TEST(setrlimit(RLIMIT_NOFILE, tc->rlimt));
> if (TST_RET != -1) {
> - tst_res(TFAIL, "call succeeded unexpectedly");
> + tst_res(TFAIL, "call succeeded unexpectedly "
> + "(nr_open=%u rlim_cur=%lu rlim_max=%lu)", nr_open,
> + (unsigned long)(tc->rlimt->rlim_cur),
> + (unsigned long)(tc->rlimt->rlim_max));
> return;
> }
>
> @@ -65,10 +69,16 @@ static void verify_setrlimit(unsigned int n)
>
> static void setup(void)
> {
> + if (tst_kvercmp(2, 6, 25) < 0) {
> + nr_open = NR_OPEN;
> + } else {
> + SAFE_FILE_SCANF("/proc/sys/fs/nr_open", "%u", &nr_open);
> + }
Can we do here without the explicit kernel version check?
What about:
#define NR_OPEN_PATH "/proc/sys/fs/nr_open"
...
if (acess(NR_OPEN_PATH, F_OK))
SAFE_FILE_SCANF(NR_OPEN_APTH, "%u", &nr_open);
Otherwise it looks good.
--
Cyril Hrubis
chrubis@suse.cz
^ permalink raw reply [flat|nested] 3+ messages in thread
* [LTP] [PATCH] syscalls/setrlimit03.c: read /proc/sys/fs/nr_open for RLIMIT_NOFILE limit
2019-01-30 16:15 ` Cyril Hrubis
@ 2019-01-30 16:22 ` Tommi Rantala
0 siblings, 0 replies; 3+ messages in thread
From: Tommi Rantala @ 2019-01-30 16:22 UTC (permalink / raw
To: ltp
ke 30. tammik. 2019 klo 18.15 Cyril Hrubis (chrubis@suse.cz) kirjoitti:
> > static void setup(void)
> > {
> > + if (tst_kvercmp(2, 6, 25) < 0) {
> > + nr_open = NR_OPEN;
> > + } else {
> > + SAFE_FILE_SCANF("/proc/sys/fs/nr_open", "%u", &nr_open);
> > + }
>
> Can we do here without the explicit kernel version check?
>
> What about:
>
> #define NR_OPEN_PATH "/proc/sys/fs/nr_open"
>
> ...
>
> if (acess(NR_OPEN_PATH, F_OK))
> SAFE_FILE_SCANF(NR_OPEN_APTH, "%u", &nr_open);
Alright, that should work too. I'll send patch v2 tomorrow.
-Tommi
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2019-01-30 16:22 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2019-01-30 15:00 [LTP] [PATCH] syscalls/setrlimit03.c: read /proc/sys/fs/nr_open for RLIMIT_NOFILE limit Tommi Rantala
2019-01-30 16:15 ` Cyril Hrubis
2019-01-30 16:22 ` Tommi Rantala
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.