Catch up with -Linus, resync with the Reiserfs folks, and
fix up some of the pending problems with the recently merged stuff.
Plus the usual mixed bag of one liners.
Patch against 2.5.2 vanilla is available from:
http://www.codemonkey.org.uk/patches/2.5/patch-2.5.2-dj7.diff.gz
When kernel.org's upload master returns, it'll also be at
ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.5/
-- Davej.
2.5.2-dj7
o Merge 2.5.3pre6
o Remove fs.h inclusion from sched.h again. (Christoph Hellwig)
o Remove some segment.h inclusions that reappeared. (Me)
o Unmangle dl2k crc fix from -dj6 (Jim McDonald)
o Fix tsdev compile. (Me)
o aty128fb & radeonfb compile fixes. (James Simmons)
o Updated Config.help entries for input layer. (Vojtech Pavlik)
o Input layer tweak for old IBM keyboards. (Vojtech Pavlik)
o Fix USB HID feature report output. (Vojtech Pavlik)
o Workaround some broken PS/2 mice. (Vojtech Pavlik)
o Don't filter outgoing fields to HID defined ranges. (Vojtech Pavlik)
o Disable address in scatterlist for sg. (Douglas Gilbert)
o Limit NR_IRQS in no IO-APIC case. (Brian Gerst)
o Sonypi driver update (C1MRX Vaio). (Stelian Pop)
o Remove bogus release_region in eexpress. (Gianluca Anzolin)
o Neofb compile fixes. (James Simmons)
o Reiserfs update. (all@namesys)
o Further reiserfs fixes. (Oleg Drokin)
o Fix keyboard not working with nothing in AUX port. (Vojtech Pavlik)
o Small devfs changes. (Richard Gooch)
o Rage128 Pro TF identification to aty128fb. (James Simmons)
2.5.2-dj6
o Merge 2.5.3pre5
o Merge 2.4.18pre7
o Fix pci_unmap_addr_SET typo. (David S. Miller)
o Make dl2k driver use generic crc function. (Jim McDonald)
o Revoke 007's right to kill. (John Levon)
o Drop i2oblock changes on Alans request.
o removes gcc-2.4.5 workaround. (Adrian Bunk)
o Further input/usb updates. (Vojtech Pavlik)
o Trigraph warning cleanup for wavelan_cs. (Jean Tourrilhes)
o driverfs support for USB. (Greg KH)
o ext2 FS corruption fix. (Alexander Viro)
o Work around reiserfs kmem_cache_create() problem. (Andi Kleen)
o CONFIG_MELAN typo in i386/setup.c (Robert Schwebel)
o Scheduler update to J7. (Ingo Molnar)
o Some Config.help / Config.in cleaning. (Me)
o Reiserfs update. (Oleg Drokin et al)
| Oleg, not all of this applied. I'll do the rest by
| hand next time.
o Convert PnPBIOS to use new subsys_initcall() (Me)
o Move screen_base to struct fb_info (James Simmons)
o Preliminary EISA support for driverfs. (Me)
| don't take this too seriously yet, it needs work.
o UFS blocksize fix. (Zwane Mwaikambo)
2.5.2-dj5
o Merge 2.5.3-pre3
o Merge 2.4.18pre6
o ieee1394 videodev compile fix. (Frank Davis)
o Removed bogus CVSIDs (Eagle eye Christoph Hellwig)
o Register EISA io ports. (Paul Gortmaker, me)
o Various APM tweaking. (Stephen Rothwell, others)
o Skeleton PCI hotplug driver. (Greg KH)
o Make IMM driver work again. (Rich Baum)
o PPC idleloop bugfix. (Rusty Russell)
o Fix USB mouse modular compile. (Greg KH)
o Make fbgen compile modular. (Various)
o Fix SMP kernel on UP boot. (Al Viro)
o Plug usb_make_path memory leaks (Greg KH)
o Only incr io_count on bio_alloc success. (Badari Pulavarty)
o Fix block backed loop mounts. (Adam Richter)
o DAC960 bio changes. (Jim McDonald)
o i2o block bio changes. (Jim McDonald)
2.5.2-dj4
o Merge 2.5.3-pre2
o Scheduler update to J4. (Ingo Molnar)
o Input layer updates. (Vojtech Pavlik, James Simmons)
o Matroxfb Configure.help updates. (Robert love)
o Framebuffer colourmap improvements. (James Simmons)
o Fix FAT infinite loops. (OGAWA Hirofumi)
o Various compile fixes. (Me)
o Small cpqarray cleanup. (Me)
o devfs fixes. (Richard Gooch)
| should fix the no-booting problem some people saw.
o ipv6 endian fixes. (Russell King)
o Make AFFS fsx-proof. (Roman Zippel)
2.5.2-dj3
o Remove/Add some mismerged bits. (Me)
o Reiserfs rename fixes. (Oleg Drokin)
o Remove 2.4 only netdriver changes. (Jeff Garzik, Me)
o Scheduler update to J2 (Ingo Molnar)
o GUID partition support update. (Matt Domsch)
o Configure help entries for IDE. (Andre Hedrick, Rob Radez,
Anton Altaparmakov)
o Reduce NTFS vmalloc use. (Anton Altaparmakov)
o Parallel port SCSI zip driver update. ([email protected])
o Iforce & Vortex joystick compile fix. (James Simmons)
o IDE Tape driver bio fixes. (Frank Davis)
o i820 & i830mp AGPGart & APM fix. (Nicolas Aspert)
o i820up AGPGart recognition. (Daniele Venzano)
o Radeonfb 1400x1050 mode timings. (Michael Clark)
2.5.2-dj2
o Merge 2.4.18pre4
o Remove duplicate soundblaster ISAPNP ID. (Jeff Garzik)
o devexit fix for dmfe. (Jeff Garzik)
o Multiport tulip irq assignment fix. (Christoph Dworzak)
o Small include file cleanup. (Andi Kleen)
o message cleanup of fatfs (OGAWA Hirofumi)
o GUID Partition Tables support. (Matt Domsch)
o Hyperthreading support for MTRR. (Sunil Saxena)
o More fbdev infrastructure work. (James Simmons)
o Numerous advansys driver fixes. (Douglas Gilbert)
o buffer.c thinko. (Andrew Morton)
o Fix ramdisk compile breakage. (Me)
o Fix acpitable.c mapping problems. (James Cleverdon)
o Input layer reworking. (James Simmons,
Vojtech Pavlik)
o Netfilter build fix. (Steven Cole)
o ATA PIO & Multimode fixes. (Jens Axboe)
o Update scheduler to J0. (Ingo Molnar, others)
o fbdev colormap cleanup. (James Simmons)
2.5.2-dj1
o Merge 2.5.2 final.
o Merge 2.5.3pre1
o Numerous compile fixes. (Various)
o Fix crc32 JFFS2 problem. (Russell King)
o Remove left over ARM bits from 2.5.1-dj15. (Me)
o Mips Magnum fb compile fix. (Me)
o Update to sched-I3 (Ingo Molnar)
o Add missing cp1250 file. (Me)
2.5.1-dj15
o Merge selective bits of 2.4.18pre3ac1 & ac2
| Drop rmap (except for rate-limit oom_kill change),
| IDE changes & 32bit uid quota
o Add 'nowayout' module param for watchdogs. (Matt Domsch)
o BSD partition fixes. (Andries Brouwer)
o wavelan_cs update (Jean Tourrilhes)
o Numerous LVM fixes. (andersg)
o Prevent ramdisk buffercache corruption. (Andrea Arcangeli)
o MS_ASYNC implementation. (Andrea, Andrew Morton)
o Truncate blocks when prepare_write() fails. (Andrea, Andrew Morton)
o winbond-840 OOM handling. (Manfred Spraul)
o Natsemi OOM handling. (Manfred Spraul)
o Eliminate some stalls in i386 syscall path. (Alex Khripin)
o Export release_console_sem() (Andrew Morton)
o Remove bogus sbp2 changes. (Christoph Hellwig)
o Remove i386 mmu_context.h (Me)
o Remove reiserfs build warnings. (Me)
o Fix ignorance of SCSI I/O errors. (Peter Osterlund)
o Fix IDE floppy thinko. (Luc Van Oostenryck)
o Radeonfb compile fixes. (Erik Andersen)
o Radeonfb flat panel support. (Michael Clark)
o Remove bogus extraneous return. (Paul Gortmaker)
o Fix potential oom-killer race. (Andres Salomon)
o Fix bio + highmem bounce BUG(). (Jens Axboe)
o PATH_MAX fixes. (Rusty Russell)
o Frame buffer _setcolreg changes. (James Simmons)
--
Dave Jones. http://www.codemonkey.org.uk
SuSE Labs.
> 2.5.2-dj7
> o Workaround some broken PS/2 mice. (Vojtech Pavlik)
What is this about exactly?
--
Balazs Pozsar
On Wed, Jan 30, 2002 at 12:19:27AM +0100, Pozsar Balazs wrote:
> > 2.5.2-dj7
> > o Workaround some broken PS/2 mice. (Vojtech Pavlik)
> What is this about exactly?
Some PS2 mice forget to ACK the GetID command before sending
a response.
--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs
> > > 2.5.2-dj7
> > > o Workaround some broken PS/2 mice. (Vojtech Pavlik)
> > What is this about exactly?
>
> Some PS2 mice forget to ACK the GetID command before sending
> a response.
I just asked, because sometimes gpm can lock up my keyboard if it cannot
read psaux. (it's 2.4) Might it be related?
--
pozsy
> > > > 2.5.2-dj7
> > > > o Workaround some broken PS/2 mice. (Vojtech Pavlik)
> > > What is this about exactly?
> >
> > Some PS2 mice forget to ACK the GetID command before sending
> > a response.
>
> I just asked, because sometimes gpm can lock up my keyboard if it cannot
> read psaux. (it's 2.4) Might it be related?
Have you tried the DJ tree with the new input devices to see if you still
have this problem?
P.S
In dmi_scan.c this is a hook to deal with the PS/2 mouse on Dell
Latitude C600. Can someone with this machine test the new input drivers on
it. I like to see if we need some kind of fix for this device.
> In dmi_scan.c this is a hook to deal with the PS/2 mouse on Dell
> Latitude C600. Can someone with this machine test the new input drivers on
> it. I like to see if we need some kind of fix for this device.
You I suspect will. When the machine resumes it likes to re-enable the mouse
pad irrespective of whether it is being used - so you get an IRQ12. Even
more fun if you ignore that IRQ you dont get keyboard events because the
microcontroller (or SMM code impersonating it - who knows these days) is
waiting for the ps/2 event to be handled first.
The alternative (possibly cleaner) fix on those machines would be to turn
the PS/2 port on always and process/discard output if its not wanted by
the user
> > I just asked, because sometimes gpm can lock up my keyboard if it cannot
> > read psaux. (it's 2.4) Might it be related?
>
> Have you tried the DJ tree with the new input devices to see if you still
> have this problem?
I only used 2.4, but i'll give it a try next time.
--
Pozsar Balazs
> > In dmi_scan.c there is a hook to deal with the PS/2 mouse on Dell
> > Latitude C600. Can someone with this machine test the new input drivers on
> > it. I like to see if we need some kind of fix for this device.
>
> You I suspect will. When the machine resumes it likes to re-enable the mouse
> pad irrespective of whether it is being used - so you get an IRQ12. Even
> more fun if you ignore that IRQ you dont get keyboard events because the
> microcontroller (or SMM code impersonating it - who knows these days) is
> waiting for the ps/2 event to be handled first.
Oh man is that brain dead.
> The alternative (possibly cleaner) fix on those machines would be to turn
> the PS/2 port on always and process/discard output if its not wanted by
> the user
This could be easily arranged with the new input drivers with it modular
design. Since for the ix86 platform most people will want PS/2 input
support to be built in. The only expection are the USB only users. I guess
with the Dell Latitude C600 we will have to force i8042.c to be built in.
Vojtech what do you think about this solution?
On Wed, Jan 30, 2002 at 10:22:02AM -0800, James Simmons wrote:
> > > In dmi_scan.c there is a hook to deal with the PS/2 mouse on Dell
> > > Latitude C600. Can someone with this machine test the new input drivers on
> > > it. I like to see if we need some kind of fix for this device.
> >
> > You I suspect will. When the machine resumes it likes to re-enable the mouse
> > pad irrespective of whether it is being used - so you get an IRQ12. Even
> > more fun if you ignore that IRQ you dont get keyboard events because the
> > microcontroller (or SMM code impersonating it - who knows these days) is
> > waiting for the ps/2 event to be handled first.
>
> Oh man is that brain dead.
The i8042 has a single byte output buffer shared by both the keyboard
and a mouse. If it's full, no more data is accepted from the keyboard or
mouse. Hence the problem above.
> > The alternative (possibly cleaner) fix on those machines would be to turn
> > the PS/2 port on always and process/discard output if its not wanted by
> > the user
>
> This could be easily arranged with the new input drivers with it modular
> design. Since for the ix86 platform most people will want PS/2 input
> support to be built in. The only expection are the USB only users. I guess
> with the Dell Latitude C600 we will have to force i8042.c to be built in.
> Vojtech what do you think about this solution?
I don't think we need to have it built in. If we need keyboard support
(which is likely), we'll have it, and if we don't need keyboard, then
we can safely ignore the IRQ12 as well.
And i8042.c, once power management is implemented in it, will reset the
keyboard controller and flush its buffers upon resume from sleep anyway.
(A problem may arise if the machine doesn't tell us about sleep/wake and
handles it all in SMM ...)
--
Vojtech Pavlik
SuSE Labs