Mailing List archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[vdr] Re: vdr-1.3.13 with NTPL



Alexw schrieb:
Hi there,

I confirm that NTPL is running fine. I was using VDR v1.3.12 with NTPL for weeks without any issue. Now I am running v1.3.13 with the same patch since the new version was released and the behavior of the system is normal. FYI I tried with kernel version 2.6.9-rc1, 2.6.9 and 2.6.9 + linuxtv latest cvs on a TT-FF and a budget-ci.

Alex

On Thursday 21 October 2004 11:08, Klaus Schmidinger wrote:

Alfred Zastrow wrote:

Hi,

since I don't have NTPL-less systems anymore, I'm forced to run vdr with
NTPL.

My vdr-build-system in short:

linux-2.6.8.1
glibc-2.3.4-20040701 (tarball is from Linuxfromscratch)
gcc-3.4.2
binutils-2.15

With the folloing two changes above (commenting out the NTPL-exit in
vdr.c and the usleep in dvb-player.c) vdr runs flawless with NTPL.
Also jumping between cutting marks appears normal.
Right now I'm not able to meassure the cpu-load, but in my case
(dedicated system) this is not an issue.

--- vdr.orig.c  Sun Oct 17 13:50:21 2004
+++ vdr.c       Wed Oct 20 21:40:39 2004
@@ -91,7 +91,7 @@
   if (confstr(_CS_GNU_LIBPTHREAD_VERSION, LibPthreadVersion,
sizeof(LibPthreadVersion)) > 0) {
      if (strstr(LibPthreadVersion, "NPTL")) {
         fprintf(stderr, "vdr: please turn off NPTL by setting 'export
LD_ASSUME_KERNEL=2.4.1' before starting VDR\n");
-        return 2;
+//        return 2;
         }
      }
 #endif

--- dvbplayer.orig.c    Fri Oct 15 15:07:55 2004
+++ dvbplayer.c Thu Oct 21 08:11:28 2004
@@ -490,8 +490,8 @@
                     break;
                     }
                  }
-              else//XXX
-                 usleep(1); // this keeps the CPU load low
+//              else//XXX
+//                 usleep(1); // this keeps the CPU load low
               }

            // Store the frame in the buffer:

regards
Alfred
Great!

I'll take it from here and will see what can be done (if at all necessary)
to keep the CPU load low without that usleep().

So, if this actually fixes all NPTL trouble, I guess it would be ok to
remove the NPTL check altogether. Can others please check whether they can
confirm Alfred's findings?

Klaus


It works fine with NTPL under kernel 2.6.8-r10 (gentoo-dev-sources) and glibc 2.3.4.20040808-r1

--
this mail was sent using 100% recycled electrons




Home | Main Index | Thread Index