Hi,
--- On Fri, 12/12/08, Vinay Sridhar <[email protected]> wrote: > From: Vinay Sridhar <[email protected]> > Subject: Re: [LTP] [PATCH] Stime01: Use Gettimeofday() Instead of Time() > To: [email protected] > Cc: [email protected], [email protected] > Date: Friday, December 12, 2008, 2:55 PM > Rishi, > > The stime01 test case actually exposed a kernel bug. While > calling > sys_stime, there is an intra-tick correction that is > compensated in the > gettimeofday path but not handled in the sys_time path. > However, the above commit modifies some of the > timekeeping.c functions, > including do_settimeofday and has fixed the issue. On the > latest kernel, > the stime01 test passes. So I dont suggest any changes to > the test. > Thanks for pointing out. So, the upstream kernel is behaving the same without this patch -- off-by-one error by time() and on x86-64 only? If so, that looks like a good candidate to backport. CAI Qian > Thanks, > Vinay > > > > On Fri, 2008-12-12 at 10:29 +0530, Rishikesh K. Rajak > wrote: > > On Thu, 2008-12-11 at 20:32 -0800, CAI Qian wrote: > > > Hi, > > > > > > Settimeofday()/gettimeofday() and time() are > known not to keep > > > consistent time against one another in RHEL 5 > kernel[1]. As the > > > result, stime01 test case failed because the date > of time() returned > > > was less than the date that stime() > (settimeofday() on x86-64) set. > > > > > > # uname -ra > > > > > > # ./stime01 > > > stime01 1 FAIL : stime() fails to set > system's time > > > > Hi Vinay, > > > > As per our discussion i feel it is a kernel problem > not ltp issue. > > can you clarify here? Is this commit is working fine > for you ? > > > > > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=9a055117d3d9cb562f83f8d4cd88772761f4cab0 > > > > I feel you have tested it and it works fine for you. > Then i think > > there is no requirement of ltp testcase modification > here if it is a > > kernel issue. > > > > -Rishi > > > > > > > > Some debug information: > > > curr_time = 1228827598 > > > pres_time = 1228827607 > > > > > > We can see that "pres_time != curr_time + > 10". > > > > > > [1] > https://bugzilla.redhat.com/show_bug.cgi?id=475477 > > > > > > Signed-off-by: CAI Qian <[email protected]> > > > > > > --- > testcases/kernel/syscalls/stime/stime01.c.orig 2008-12-12 > 11:25:49.952135991 +0800 > > > +++ > testcases/kernel/syscalls/stime/stime01.c 2008-12-12 > 11:42:43.456151307 +0800 > > > @@ -73,6 +73,7 @@ > > > #include <time.h> > > > #include <string.h> > > > #include <sys/stat.h> > > > +#include <sys/time.h> > > > #include <signal.h> > > > > > > #include "test.h" > > > @@ -83,9 +84,8 @@ > > > char *TCID="stime01"; /* Test program > identifier. */ > > > int TST_TOTAL=1; /* Total number of test cases. > */ > > > extern int Tst_count; /* Test Case counter for > tst_* routines */ > > > -time_t curr_time; /* system's current time > in seconds */ > > > time_t new_time; /* system's new time */ > > > -time_t tloc; /* argument var. for time() */ > > > +struct timeval tv; /* argument var. for > gettimeofday() */ > > > int exp_enos[]={0}; > > > > > > void setup(); /* Main setup function of test > */ > > > @@ -96,8 +96,7 @@ > > > { > > > int lc; /* loop counter */ > > > char *msg; /* message returned from parse_opts > */ > > > - time_t pres_time; /* system's present time > */ > > > - > > > + > > > /* Parse standard options given to run the > test. */ > > > msg = parse_opts(ac, av, (option_t *) NULL, > NULL); > > > if (msg != (char *) NULL) { > > > @@ -138,7 +137,7 @@ > > > * Get the system's current time after > call > > > * to stime(). > > > */ > > > - if ((pres_time = time(&tloc)) < 0) { > > > + if ((gettimeofday(&tv, NULL)) < 0) { > > > tst_brkm(TFAIL, cleanup, > > > "time() failed to get " > > > "system's time after stime, > " > > > @@ -147,8 +146,8 @@ > > > } > > > > > > /* Now do the actual verification */ > > > - if ((pres_time != new_time) && > > > - (pres_time != new_time + 1)) { > > > + if ((tv.tv_sec != new_time) && > > > + (tv.tv_sec != new_time + 1)) { > > > tst_resm(TFAIL, "stime() fails to set > " > > > "system's time"); > > > } else { > > > @@ -194,15 +193,15 @@ > > > TEST_PAUSE; > > > > > > /* Get the current time */ > > > - if ((curr_time = time(&tloc)) < 0) { > > > + if ((gettimeofday(&tv, NULL)) < 0) { > > > tst_brkm(TBROK, cleanup, > > > - "time() failed to get current time, > errno=%d", > > > + "gettimeofday() failed to get current > time, errno=%d", > > > errno); > > > /*NOTREACHED*/ > > > } > > > > > > /* Get the system's new time */ > > > - new_time = curr_time + INCR_TIME; > > > + new_time = tv.tv_sec + INCR_TIME; > > > } /* End setup() */ > > > > > > /* > > > > > > > > > > ------------------------------------------------------------------------------ > > > SF.Net email is Sponsored by MIX09, March 18-20, > 2009 in Las Vegas, Nevada. > > > The future of the web can't happen without > you. Join us at MIX09 to help > > > pave the way to the Next Web now. Learn more and > register at > > > > http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ > > > _______________________________________________ > > > Ltp-list mailing list > > > [email protected] > > > > https://lists.sourceforge.net/lists/listinfo/ltp-list ------------------------------------------------------------------------------ SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ _______________________________________________ Ltp-list mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ltp-list
