> Quoting Dotan Barak <[EMAIL PROTECTED]>: > Subject: Question about registering the [vdso] memory section in user level > > Hi. > > In our regression, there is a test case in which we register the last > VMA of a process that has a read permission. > Only in kernel 2.6.20-rc5 i get a failure. > > Here is the last line that i got from executing "cat /proc/1873/maps": > > > In kernel 2.6.16.21-0.8-smp: > 2b42d103f000-2b42d1116000 r--p 00000000 08:07 647899 > /usr/lib/locale/en_US.utf8/LC_COLLATE > 2b42d1116000-2b42d1117000 r--p 00000000 08:07 633217 > /usr/lib/locale/en_US.utf8/LC_TIME > 2b42d1117000-2b42d1118000 r--p 00000000 08:07 647880 > /usr/lib/locale/en_US.utf8/LC_NUMERIC > 2b42d1118000-2b42d114b000 r--p 00000000 08:07 647898 > /usr/lib/locale/en_US.utf8/LC_CTYPE > 2b42d114b000-2b42d114c000 rw-p 2b42d114b000 00:00 0 > 7fffda251000-7fffda267000 rw-p 7fffda251000 00:00 0 > [stack] > ffffffffff600000-ffffffffffe00000 ---p 00000000 00:00 0 > [vdso] > > In kernel 2.6.20-rc5: > 2ba22fd99000-2ba22fd9b000 rw-p 0000a000 08:03 309500 > /lib64/libnss_files-2.3.4.so > 2ba22fd9b000-2ba232be6000 r--p 00000000 08:03 68735 > /usr/lib/locale/locale-archive > 2ba232be6000-2ba232bec000 r--s 00000000 08:03 97940 > /usr/lib64/gconv/gconv-modules.cache > 2ba232bec000-2ba232bee000 rw-p 2ba232bec000 00:00 0 > 7fff7ae2d000-7fff7ae43000 rw-p 7fff7ae2d000 00:00 0 > [stack] > ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 > [vdso] > > It seems that in kernel 2.6.20-rc5 the last VMA which has a read > permission is the [vdso] section but when i try > to register it i get a failure. > > Is it wrong to register this section?
Yes, you can't DMA to VDSO VMA I don't think. -- MST _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
