Post by Andrea Arcangel » Thu, 25 Jul 2002 03:00:12




Only in 2.4.19rc3aa1: 00_net-softirq-1

        Lower AF_UNIX(/and similar local tranfers) latency.

Only in 2.4.19rc2aa1: 00_poll-speedup-1
Only in 2.4.19rc3aa1: 00_poll-speedup-2

        Return exporting a few methods (get_fd_set/zero_fd_set).

Only in 2.4.19rc3aa1: 00_relax-timer-sync-1

        Use rep;nop while waiting a timer to go away from the cpu.

Only in 2.4.19rc3aa1: 00_sched-O1-aa-2.4.19rc3-1.gz
Only in 2.4.19rc2aa1: 00_sched-O1-rml-2.4.19-pre9-2.gz
Only in 2.4.19rc2aa1: 02_sched-19pre8ac5-1
Only in 2.4.19rc2aa1: 02_sched-alpha-1
Only in 2.4.19rc2aa1: 02_sched-sparc64-1
Only in 2.4.19rc2aa1: 02_sched-x86-1
Only in 2.4.19rc2aa1: 10_o1-sched-64-cpu-1
Only in 2.4.19rc2aa1: 10_o1-sched-fixes-2
Only in 2.4.19rc2aa1: 10_o1-sched-updates-A4-3

        Synchronized with sched-2.4.19-rc2-A3 from Ingo, I didn't merge
        the new features, just the bugfixes. The major ones
        are my proposed change to sched_yield to just move sched_yield
        tasks directly into the expired queues that should fix
        sched_yield related hangs like with tomcat during java
        garbage collection, and various important performance fixes like the
        load_balancing that was completely broken in the previous
        implementations, and other fixes from Ingo like the wakeup_sync that
        was not honouring cpus_allowed and various minor optimizations.

Only in 2.4.19rc2aa1: 08_qlogicfc-template-aa-2
Only in 2.4.19rc3aa1: 08_qlogicfc-template-aa-3

        Deleted one more line during merging.

Only in 2.4.19rc2aa1: 10_parent-timeslice-10

        Now part of O1 sched, except Ingo apparently forgot a place
        in exit.c, that is included into 00_sched-O1-aa-2.4.19rc3-1.gz
        (with that fix added, it's completely equivalent to the previous

Only in 2.4.19rc3aa1: 50_uml-patch-2.4.18-41.gz
Only in 2.4.19rc3aa1: 50_uml-patch-2.4.18-42.gz

        Latest updates from Jeff.

Only in 2.4.19rc2aa1: 51_uml-o1-1
Only in 2.4.19rc3aa1: 51_uml-o1-2

        Merged some new bit from the 2.5 uml port to fix o1 sched hangs under

Only in 2.4.19rc2aa1: 88_x86_64-poll-1

        Return using the exported symbols.

Only in 2.4.19rc2aa1: 90_buddyinfo-1
Only in 2.4.19rc3aa1: 90_buddyinfo-2

        New implementation from William Lee Irwin III.

Only in 2.4.19rc2aa1: 90_proc-mapped-base-1
Only in 2.4.19rc3aa1: 90_proc-mapped-base-2

        Show /proc/<pid>/mapped_base only on archs where it is significant.
        Enabled for x86 and s390 only so far. Wonder what to do on the 32/64
        bit archs like x86-64, s390x, ppc64, ia64, sparc64 where I cannot
        generally inherit the mapped base from a 64bit parent to a 32bit child.
        I guess it's ok that the /proc/<pid>/mapped_base isn't there so in
        32bit compatility mode the app will just use the default, since the
        whole point of those archs is to run those user apps in 64bit mode
        where mapped_base won't need to be dynamic.

Only in 2.4.19rc3aa1: 90_s390-aa-1

        Make it compile on s390.

Only in 2.4.19rc2aa1: 94_discontigmem-meminfo-1
Only in 2.4.19rc3aa1: 94_discontigmem-meminfo-2

        Fix meminfo node reporting.

Only in 2.4.19rc2aa1: 96_inode_read_write-atomic-1
Only in 2.4.19rc3aa1: 96_inode_read_write-atomic-2

        Use two sequence numbers for 486/386 SMP compiles and in general
        for SMP compiles of 32bit archs not providing (or not yet implementing :)
        64bit atomic get/set. (also renamed read_64bit to get_64bit for
        symmetry with set_64bit) Added some comment about preempt significant
        only for the 2.5 port of it.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/


1. 2.4.19rc2 -> 2.4.19rc3 : no more eth

Dear kernel crew.

I follow the stable kernel and run 2.4.18 just fine on 3 machines.

Recently I thought about trying 2.4.19rc kernels.  That went fine as
well with one exception.

One machine (with 2 ethernet controllers)
 Intel Corp. 82557/8/9 [Ethernet Pro 100] (rev 0c)
 3Com Corporation 3c900B-TPO [Etherlink XL TPO] (rev 04)
will no longer recognise -any- of those.

Note that this 'artifact' only happens with 2.4.19rc3.  The .config
between 2.4.19rc2 did not change 2.4.19rc3

During boot I see OOPses (which for some reason, my syslog does not

Because this is a rather severe change [:-)] I wonder if I this rings
a bell.

Thanks for your valuable time


Marc Duponcheel     Multicast Development Engineer      Cisco Systems

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

2. Hardware related code in Linux-kernel

3. Kernel patching 2.4.19pre1 -> 2.4.19pre2

4. Cable Modem & Windows NT Speed Tips

5. 2.4.19-rc2 -> 2.4.19-rc3 : no more eth (fwd)

6. Refernced Symbol Not Found

7. 3Ware ok 2.4.19, dies 2.4.19-ac4

8. PCI PNP Modem under Linux

9. Promise 20267 hangs with 2.4.19-pre3 and 2.4.19-pre3-ac3

10. 2.4.19-pre6aa1 (possible all kernel after 2.4.19-pre2) athlon PCI workaround

11. Kernel panic 2.4.19-pre6 AND 2.4.19-pre5-ac3 - More info - ksymoops

12. kbuild25 version 3.0 for 2.4.19-pre9 and 2.4.19-pre9-ac3

13. fix for pci_enable_device in 2.4.19ac3