[parisc-linux] 2.5 status update 2003-02-11

Matthew Wilcox willy@debian.org
Tue, 11 Feb 2003 23:03:54 +0000

Warning: Linus' tree will probably now build for PA-RISC.  This does
NOT mean that it will work.  You still need to pick up a PA patch from
ftp.parisc-linux.org for it to work.

The good news is that patch is now smaller than ever.  The .gz is down
to 50k; uncompressed it's around 200k.  Almost all of this is the
HIL drivers.

Things which've been done since the last report:
 - Start of module support (Randolph)
 - Merged 2.5.59 (James)
 - More compat chanages (Randolph / Stephen)
 - Convert lasi 82596 driver to DMA api (James)
 - Eliminate ccio_get_fake (James)
 - Raw keyboard support (Helge)
 - Merged 2.5.60 (Matthew / James)

Current status:
 - Merged to 2.5.60, boots on 712
 - 2.5.58 booted on A500, 715/64
 - 2.5.57 booted on C160
 - SuckyIO serial problems still exist.
 - Don't know what other patches need to be brought forward from 2.4
 - SMP compiles, but doesn't work yet
 - No preemptible support

Todo list:
 - review diff vs 2.4 for additional changes
 - Serial code not working for c3000
 - the fix for do_fork needs checking
 - our PDC early debug console hacks need to be cleaned up somehow
 - ad1889 driver: only works with .wav;  Convert to ALSA (ggg, tausq, tbone)
 - ns87415 dma doesn't work reliably on suckyio-systems
 - CPU hotplug support (useful to take flaky CPU "offline")
 - flush_tlb_kernel_range is horribly inefficient
 - flush_icache_user_range could flush just that range rather than the
   entire page.
 - (ab)use kmap/kunmap on 64-bit to eliminate flush_dcache calls.
 - cp_new_stat32 for sys_parisc32.c is inefficient; maybe it's better
   to fill in a tmp stat32 and just do copy_to_user in one go at the end?
 - syscall signal return path needs work
 - changes in arch/parisc/kernel/signal.c regarding HACK needs testing
 - signal.c #if CACHE_FLUSHING_IS_NOT_BROKEN needs to be solved
 - task_struct/thread_info split -- task_struct should not be visible in
   entry.S, we need to move some items into thread_info -- this includes
   pt_regs and maybe some of the flags (ptrace, etc)
 - Use zone_highmem for the 256MB of RAM at 64GB on Astro-based machines.
   (NB, what does Stretch do?)
 - investigate not putting in extable entries for put_kernel_asm; will
   probably reduce kernel size
 - call to pdc_suspend_usb() in inventory.c crashes c3k with kernel 2.5
 - fix HIL problem: ksoftirqd/0 eats 56% cpu (kernel 2.4 & kernel 2.5)
 - write floppy driver for lasi
 - write floppy driver for suckyio
 - XFS goes into an infinite loop when mounting volume.
 - spifi driver (rbrad)
 - write optimized versions of lcopy_{to,from}_user. Can we port the
   hpux versions?
 - kernel module loader support
 - IRQ cpu-affinity support
 - NPTL kernel support (CLONE_*TID flags need to be correctly handled by 
   sys_clone() and friends)
 - port sym53c8xx hppa changes forward (still needed?)

In progress:
 - Some options still need Kconfig help entries (see below!)
 - Look at Config.help too. (tbone, HIL -> helge?)
 - commit beta ALSA harmony driver (laurent)
 - port hil_kbd.c to new input layer
 - port hil_ptr.c to new input layer
 - oprofile support (tausq)

"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk