I got an Error while compiling 2.6.18-rc3-mm2:
AR arch/i386/lib/lib.a
GEN .version
CHK include/linux/compile.h
UPD include/linux/compile.h
CC init/version.o
LD init/built-in.o
LD .tmp_vmlinux1
kernel/built-in.o(.text+0x45667): In function `bacct_add_tsk':
include/linux/time.h:130: undefined reference to `__divdi3'
make: *** [.tmp_vmlinux1] Error 1
I attached the .config file.
Have a good day.
Matt
On Sun, 6 Aug 2006 23:30:19 -0400
Matt Reuther <[email protected]> wrote:
> I got an Error while compiling 2.6.18-rc3-mm2:
>
> AR arch/i386/lib/lib.a
> GEN .version
> CHK include/linux/compile.h
> UPD include/linux/compile.h
> CC init/version.o
> LD init/built-in.o
> LD .tmp_vmlinux1
> kernel/built-in.o(.text+0x45667): In function `bacct_add_tsk':
> include/linux/time.h:130: undefined reference to `__divdi3'
> make: *** [.tmp_vmlinux1] Error 1
>
> I attached the .config file.
>
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc3/2.6.18-rc3-mm2/hot-fixes/csa-basic-accounting-over-taskstats-fix.patch
should fix this, thanks.
On Monday 07 August 2006 01:21 am, Andrew Morton wrote:
> On Sun, 6 Aug 2006 23:30:19 -0400
>
> Matt Reuther <[email protected]> wrote:
> > I got an Error while compiling 2.6.18-rc3-mm2:
> >
> > AR arch/i386/lib/lib.a
> > GEN .version
> > CHK include/linux/compile.h
> > UPD include/linux/compile.h
> > CC init/version.o
> > LD init/built-in.o
> > LD .tmp_vmlinux1
> > kernel/built-in.o(.text+0x45667): In function `bacct_add_tsk':
> > include/linux/time.h:130: undefined reference to `__divdi3'
> > make: *** [.tmp_vmlinux1] Error 1
> >
> > I attached the .config file.
>
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc3/2.
>6.18-rc3-mm2/hot-fixes/csa-basic-accounting-over-taskstats-fix.patch should
> fix this, thanks.
It does indeed fix the error. Thank you!
Matt
On Sun, Aug 06, 2006 at 10:21:29PM -0700, Andrew Morton wrote:
> On Sun, 6 Aug 2006 23:30:19 -0400
> Matt Reuther <[email protected]> wrote:
>
> > I got an Error while compiling 2.6.18-rc3-mm2:
> >
> > AR arch/i386/lib/lib.a
> > GEN .version
> > CHK include/linux/compile.h
> > UPD include/linux/compile.h
> > CC init/version.o
> > LD init/built-in.o
> > LD .tmp_vmlinux1
> > kernel/built-in.o(.text+0x45667): In function `bacct_add_tsk':
> > include/linux/time.h:130: undefined reference to `__divdi3'
> > make: *** [.tmp_vmlinux1] Error 1
> >
> > I attached the .config file.
> >
>
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc3/2.6.18-rc3-mm2/hot-fixes/csa-basic-accounting-over-taskstats-fix.patch
> should fix this, thanks.
This doesn't look correct since do_div() does not guarantee to return
more than 32bit.
What about the patch below that adds a timespec_to_us() to time.h and
uses this function in kernel/tsacct.c?
<-- snip -->
This patch adds a timespec_to_us() to include/linux/time.h and uses it
to fix a compile error in kernel/tsacct.c .
Signed-off-by: Adrian Bunk <[email protected]>
---
include/linux/time.h | 12 ++++++++++++
kernel/tsacct.c | 2 +-
2 files changed, 13 insertions(+), 1 deletion(-)
--- linux-2.6.18-rc3-mm2-full/include/linux/time.h.old 2006-08-06 19:56:50.000000000 +0200
+++ linux-2.6.18-rc3-mm2-full/include/linux/time.h 2006-08-06 20:00:51.000000000 +0200
@@ -132,6 +132,18 @@
}
/**
+ * timespec_to_us - Convert timespec to microseconds
+ * @ts: pointer to the timespec variable to be converted
+ *
+ * Returns the scalar microsecond representation of the timespec
+ * parameter.
+ */
+static inline s64 timespec_to_us(const struct timespec *ts)
+{
+ return ((s64) ts->tv_sec * USEC_PER_SEC) + ts->tv_nsec / NSEC_PER_USEC;
+}
+
+/**
* timeval_to_ns - Convert timeval to nanoseconds
* @ts: pointer to the timeval variable to be converted
*
--- linux-2.6.18-rc3-mm2-full/kernel/tsacct.c.old 2006-08-06 19:54:45.000000000 +0200
+++ linux-2.6.18-rc3-mm2-full/kernel/tsacct.c 2006-08-06 19:56:44.000000000 +0200
@@ -36,7 +36,7 @@
do_posix_clock_monotonic_gettime(&uptime);
ts = timespec_sub(uptime, current->group_leader->start_time);
/* rebase elapsed time to usec */
- stats->ac_etime = (timespec_to_ns(&ts))/NSEC_PER_USEC;
+ stats->ac_etime = timespec_to_us(&ts);
stats->ac_btime = xtime.tv_sec - ts.tv_sec;
if (thread_group_leader(tsk)) {
stats->ac_exitcode = tsk->exit_code;
On Mon, 7 Aug 2006 15:32:41 +0200
Adrian Bunk <[email protected]> wrote:
> On Sun, Aug 06, 2006 at 10:21:29PM -0700, Andrew Morton wrote:
> > On Sun, 6 Aug 2006 23:30:19 -0400
> > Matt Reuther <[email protected]> wrote:
> >
> > > I got an Error while compiling 2.6.18-rc3-mm2:
> > >
> > > AR arch/i386/lib/lib.a
> > > GEN .version
> > > CHK include/linux/compile.h
> > > UPD include/linux/compile.h
> > > CC init/version.o
> > > LD init/built-in.o
> > > LD .tmp_vmlinux1
> > > kernel/built-in.o(.text+0x45667): In function `bacct_add_tsk':
> > > include/linux/time.h:130: undefined reference to `__divdi3'
> > > make: *** [.tmp_vmlinux1] Error 1
> > >
> > > I attached the .config file.
> > >
> >
> > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc3/2.6.18-rc3-mm2/hot-fixes/csa-basic-accounting-over-taskstats-fix.patch
> > should fix this, thanks.
>
> This doesn't look correct since do_div() does not guarantee to return
> more than 32bit.
eh? We use do_div() to do 64bit/something all the time??
> What about the patch below that adds a timespec_to_us() to time.h and
> uses this function in kernel/tsacct.c?
Seems reasonable, but it'd be better as two patches..
Do we do timespec->microseconds anywhere else?
On Mon, Aug 07, 2006 at 01:24:18PM -0700, Andrew Morton wrote:
> On Mon, 7 Aug 2006 15:32:41 +0200
> Adrian Bunk <[email protected]> wrote:
>
> > On Sun, Aug 06, 2006 at 10:21:29PM -0700, Andrew Morton wrote:
> > > On Sun, 6 Aug 2006 23:30:19 -0400
> > > Matt Reuther <[email protected]> wrote:
> > >
> > > > I got an Error while compiling 2.6.18-rc3-mm2:
> > > >
> > > > AR arch/i386/lib/lib.a
> > > > GEN .version
> > > > CHK include/linux/compile.h
> > > > UPD include/linux/compile.h
> > > > CC init/version.o
> > > > LD init/built-in.o
> > > > LD .tmp_vmlinux1
> > > > kernel/built-in.o(.text+0x45667): In function `bacct_add_tsk':
> > > > include/linux/time.h:130: undefined reference to `__divdi3'
> > > > make: *** [.tmp_vmlinux1] Error 1
> > > >
> > > > I attached the .config file.
> > > >
> > >
> > > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc3/2.6.18-rc3-mm2/hot-fixes/csa-basic-accounting-over-taskstats-fix.patch
> > > should fix this, thanks.
> >
> > This doesn't look correct since do_div() does not guarantee to return
> > more than 32bit.
>
> eh? We use do_div() to do 64bit/something all the time??
Sorry, this was my thinko.
> > What about the patch below that adds a timespec_to_us() to time.h and
> > uses this function in kernel/tsacct.c?
>
> Seems reasonable, but it'd be better as two patches..
Feel free to split this patch however you want, I thought it was not
worth splitting it.
> Do we do timespec->microseconds anywhere else?
I haven't found another place.
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
Andrew Morton writes:
> On Mon, 7 Aug 2006 15:32:41 +0200
> Adrian Bunk <[email protected]> wrote:
> > This doesn't look correct since do_div() does not guarantee to return
> > more than 32bit.
>
> eh? We use do_div() to do 64bit/something all the time??
Indeed. If do_div didn't return a 64-bit quotient then
printk("%lld", ...) wouldn't work. (The remainder is 32-bit of course,
because the divisor is 32-bit.)
Paul.
On Monday 07 August 2006 09:32 am, Adrian Bunk wrote:
> On Sun, Aug 06, 2006 at 10:21:29PM -0700, Andrew Morton wrote:
> > On Sun, 6 Aug 2006 23:30:19 -0400
> What about the patch below that adds a timespec_to_us() to time.h and
> uses this function in kernel/tsacct.c?
>
>
> <-- snip -->
>
>
> This patch adds a timespec_to_us() to include/linux/time.h and uses it
> to fix a compile error in kernel/tsacct.c .
>
> Signed-off-by: Adrian Bunk <[email protected]>
>
> ---
>
> include/linux/time.h | 12 ++++++++++++
> kernel/tsacct.c | 2 +-
> 2 files changed, 13 insertions(+), 1 deletion(-)
>
> --- linux-2.6.18-rc3-mm2-full/include/linux/time.h.old 2006-08-06
> 19:56:50.000000000 +0200 +++
> linux-2.6.18-rc3-mm2-full/include/linux/time.h 2006-08-06
> 20:00:51.000000000 +0200 @@ -132,6 +132,18 @@
> }
>
> /**
> + * timespec_to_us - Convert timespec to microseconds
> + * @ts: pointer to the timespec variable to be converted
> + *
> + * Returns the scalar microsecond representation of the timespec
> + * parameter.
> + */
> +static inline s64 timespec_to_us(const struct timespec *ts)
> +{
> + return ((s64) ts->tv_sec * USEC_PER_SEC) + ts->tv_nsec / NSEC_PER_USEC;
> +}
> +
> +/**
> * timeval_to_ns - Convert timeval to nanoseconds
> * @ts: pointer to the timeval variable to be converted
> *
> --- linux-2.6.18-rc3-mm2-full/kernel/tsacct.c.old 2006-08-06
> 19:54:45.000000000 +0200 +++
> linux-2.6.18-rc3-mm2-full/kernel/tsacct.c 2006-08-06 19:56:44.000000000
> +0200 @@ -36,7 +36,7 @@
> do_posix_clock_monotonic_gettime(&uptime);
> ts = timespec_sub(uptime, current->group_leader->start_time);
> /* rebase elapsed time to usec */
> - stats->ac_etime = (timespec_to_ns(&ts))/NSEC_PER_USEC;
> + stats->ac_etime = timespec_to_us(&ts);
> stats->ac_btime = xtime.tv_sec - ts.tv_sec;
> if (thread_group_leader(tsk)) {
> stats->ac_exitcode = tsk->exit_code;
This patch also compiles for me without errors.
Matt