Hi folks,
the cleanup of my cputime patches is done and now I'm ready for the
first round of bashing ;-)
I've split the kernel patch into 5 parts, Jan works on another kernel
patch that introduces virtual cpu time slices but this isn't ready yet.
Patch number 6 is a patch against procps that make t
Hi Mark,
> No, that's not the name inside the tarball. See my note to Ulrich (or
> download the file yourself if you don't believe me). As I also said to
> Ulrich, even if that were the name of the file, I have no way of knowing
> which lines actually comprise _just_ the "multicast notifier patc
> The only significant exceptions I know of for 2.6.7 are these:
> - The shared-IPv6-card patch:
> linux-2.6.5-s390-base-14-april2004.diff
> Sadly, no one has so far come up with a solution that can be
> merged into the official kernel, so you have to apply this
> to be able to use IPv6 wit
> Martin, you are the Wizard, indeed! Thank you very much! Now i've got new
> kernel with old disks :)
Thanks for the flowers, but actually it was Conny who came up with the
idea for the fix. I "only" did the debugging part ...
blue skies,
Martin
Linux/390 Design & Development, IBM Deutschlan
> SenseID : device 0750 reports: CU Type/Mod = 3990/EC, Dev Type/Mod =
> 3390/0A (for Shark)
> SenseID : device 080b reports: CU Type/Mod = 3990/E9, Dev Type/Mod =
> 3390/0A (for Tetragon)
> SenseID : device 080c reports: CU Type/Mod = 3990/E9, Dev Type/Mod =
> 3390/0A
Hi Sergey,
> Is there a problem with "not supported by new driver" or i've just
> forgotten to compile something into the kernel?
Could you ipl with the cio_msg=yes parameter and tell me what the
"SenseID" line for the device looks like. I have the feeling that
there is a cu type/device type pair
> I don't think you're on crack,
> but the book might be. 4,64 is a UART class serial line.
> Not likely your zSeries will have that. But all HW have 5,1 console.
Well, neither is on crack. The 3215/sclp console driver use 4,64 as device
node, so the dd-book is correct. But nevertheless you ha
> This patch cleans up needless includes of asm/pgalloc.h from the
> arch/s390/ subtree. This has not been compile tested, so
> needs the architecture maintainers (or willing volunteers) to
> test.
Doesn't compile. s390_ksyms needs pgalloc.h for the definition of diag10.
The other includes of pga
Hi Sergey,
> arch/s390/kernel/traps.c: In function `show_registers':
> arch/s390/kernel/traps.c:202: inconsistent operand constraints in an `asm'
Try a newer compiler. The __get_user inline in include/asm-s390/uaccess.h
uses a "=Q" constraint for all 3.x compilers. Seems like your gcc 3.2 doesn't
Hi Stephen,
> I just happened to follow the function and noticed that on s390,
> page_test_and_clear_dirty() has the comment:
>
> * Test and clear dirty bit in storage key.
> * We can't clear the changed bit atomically. This is a potential
> * race against modification of the referenced bit. This
> I've never seen "accepted" as a state. IBM guys: Any ideas?
A dasd device is in state "accepted" after the disable ioctl. This is
done by dasdfmt, it first disables the device, then formats each track
and reenables it after writing the default partition table.
> Invalidate first track...
> da
> The patych below removes an #ifdef for kernel 2.2 from
> include/asm-s390/debug.h .
Hmm, this one somehow slipped my notice. It is the last occurence
of KERNEL_VERSION in the s390 code for 2.6. I've applied the patch
to our CVS and I will send it with the next update to Andrew.
Thanks for spotti
> Is this because the patches that are put out for 2.4.21 weren't being
> integrated into 2.4.22 by Marcello? Or is something else not happening
that
> should? Without understanding what's happening, I would think that more
of
> the patches to 2.4.21 would have been in 2.4.22.
We have quite a lo
> >The straight forward solution is to replace "__u64" with "unsigned long
long",
> >"__u32" with "unsigned int" and "__u16" with "unsigned short". Since
ptrace.h
> >is commonly used in user space (hmpf) I'll change asm-s390/ptrace.h.
>
> This doesn't really solve the problem, because there is a re
> Oh, oh. I hope I didn't start something with this. What I really should
> have said was...
>
> I haven't even tried submitting it to (who???) the original developers of
> zipl.
The original developer of zipl is Carsten Otte ([EMAIL PROTECTED]). Recently
Peter Oberparleiter ([EMAIL PROTECTED]) t
> This phenomenon appears to be caused by some interaction between the
> delivery of I/O interrupts and timer interrupts when running under
> VM (or LPAR with shared CPUs, for that matter). We haven't quite
> understood this whole issue yet, unfortunately. (Maybe Martin knows
> more details ...)
>I admit I only briefly looked at the latest code, but afaik
>especially with the 'demand timer patch' Linux uses the TOD
>to tell time, and I don't recall seeing the drift corrections
>applied. This would mean an ntpd cannot adjust time of Linux
>running on z/VM, but one might be able to change t
Hi Tim,
>Uggh, DaveM pointed out a very good issue with this fix (similar for
>Sparc64) and core files. Core files will now have truncated uid/gid
>values because fs/binfmt_elf calls NEW_TO_OLD_UID(). May be other
>places, too.
>
>I guess you should not apply this patch until I've had a better
Hi Tim,
> I am working on a patch that removes the old limit of NGROUPS.
>
> In doing so, I am checking everything that references NGROUPS. One
> such place is in arch/s390x/kernel/linux32.c.
>
> I notice these things:
> * s390x defines __kernel_old_gid_t to __kernel_gid_t (unsigned int)
> * s39
>the errors should be :
>../sysdeps/unix/sysv/linux/s390/s390-32/sysdep.S: Assembler message
>../sysdeps/unix/sysv/linux/s390/s390-32/sysdep.S:82: ERROR: backward
>reference to an unknown target "0:"
There is a bug in sysdep.S for the case non-PIC and non-reentrant. For
a "normal" compile this cas
>binutils to 2.12.1
Not good. Please use version 2.12.90.0.4 and the patch from D-Works.
blue skies,
Martin
Linux/390 Design & Development, IBM Deutschland Entwicklung GmbH
Schönaicherstr. 220, D-71032 Böblingen, Telefon: 49 - (0)7031 - 16-2247
E-Mail: [EMAIL PROTECTED]
>It's not that it's stopping because of a console event, it's just
>dropping dead at the OS level and I can't even bring it back with the
console.
Does the system drop dead without the timer patch as well? And can you do
a "#cp cpu all d p" after the system stopped and post the result?
blue skie
Hi Neale,
> Is this the same bit of logic in 2.4.7:
Indeed:
diff -urN linux-2.4.7/arch/s390/mm/fault.c linux-2.4.7-s390/arch/s390/mm/fault.c
--- linux-2.4.7/arch/s390/mm/fault.c Thu May 23 10:04:50 2002
+++ linux-2.4.7-s390/arch/s390/mm/fault.c Wed May 22 20:13:04 2002
@@ -163,6 +163,8 @@
Hi Jason,
> I've tried the attached simple program on RedHat 2.4.9-17 and
> the lastest 2.4.17 build from IBM, and both fail miserably:
This is a bug in arch/{s390,s390x}/mm/fault.c that is present since the
beginning. A small wonder that nobody noticed it until now...
The following patch should
> There might have been some historical reasons from before
> even the first public releases (which is why I asked Martin
> -- this was before I joined the team); or it might have
> been simply broken from the start.
I'd say it was broken from the start. It never caused problems
so nobody noticed.
Hi Richard,
>If you're using a 3270 console with linux-390 you may have noticed that
>when you give the "reboot" command your system typically goes into a
>disabled loop. Here's a patch to cure the problem. It should apply
>cleanly on any 2.4.7, 2.4.9, or 2.4.17 kernel tree. Alan, Martin,
>who
>The standard distribution has a few checks in console
>initialisation, which appear a bit superflous to me, and
>in fact prevent one from being able to specify conmode=
>on non vm or p390 type machines. Removing those checks
>will allow one to specify conmode=3270 and condev=1234
>as the kernel
>I don't have a clue why this would be recommended for a LINUX virtual
>machine. Does Linux get memory from CMS? It may be that IBM (or
>someone) has implemented something for similiar reasons under Linux.
The "set relpage off" setting is a leftover from the dark age. The very
first experiments
>Don't you think it would preclude having a decent console on Hercules?
If Hercules does not set the cpu id to 0xff?? (VM) or 0x7490 (P390)
then the 3215 console is disabled.
>I am going to ask Martin to kill the test for machine type
>from conmode_setup, but first someone please confirm
>So, Martin, how long does a jump take before you're on the ground
>with both feet again ;-)
About 3-4 minutes.
blue skies,
Martin
Linux/390 Design & Development, IBM Deutschland Entwicklung GmbH
Schönaicherstr. 220, D-71032 Böblingen, Telefon: 49 - (0)7031 - 16-2247
E-Mail: [EMAIL PROTECTED]
>>Do you know where I can find this timer patch ? (for 2.2.16)
>As far as I know there is such.
The timer patch is available only for 2.4.x (x = 7, 16 or 17).
>But try to reduce Linux guest size and make sure there is no
>need to page out your Linux guest.
Avoid overcommitting memory in VM. This
> I think I am doing something wrong, but I'm not sure what. I'm am trying
to
> write a very simple IOCTL module using kernel version 2.4.5 from
TurboLinux.
> I have included uaccess.h and I call the module copy_from_user. The
module
> compiles correctly, but when I do an insmod on it I get an "
101 - 132 of 132 matches
Mail list logo
