2002-03-17 01:32:43

by Dave Jones

[permalink] [raw]
Subject: Linux 2.5.6-dj1

Mostly just a 'catching up' release after my weeklong haitus.
Almost up to date with 2.4, and right up to date with 2.5 (for today at least)
This one hasn't had any testing beyond "it compiles" (testboxes are still in
packing boxes, and I'm currently a few hundred KM from home), so tread
carefully. This one is mostly just a resync-point just to start the
probably imminent 2.5.7 afresh.

Oh, and as in 2.5.7pre2, disable ACPI. It's more borken than usual.

There's still a mountain of pending bits to dig through, and lots of work
to be done in the "splitting up bits for Linus" dept.


As usual,..
Patch against 2.5.6 vanilla is available from:
ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.5/

Merged patch archive: http://www.codemonkey.org.uk/patches/merged/

Check http://www.codemonkey.org.uk/Linux-2.5.html before reporting
known bugs that are also in mainline.

-- Davej.

2.5.6-dj1
o Sync against 2.5.6
| Take my last_rx changes for hamradio over mainline.
o Merge 2.5.7pre2
o Initialise Machine check bank 0 on AMD systems. (Me)
o background checking for non-fatal machine checks. (Me)
o Fix ALSA config.in so xconfig works again. (William Stinson)
o ITE8330G IRQ router support. (Tobias Diedrich)
o Fix reiserfs oops on mount. (Oleg Drokin)
o Only offer various MIPS drivers on !MIPS arches. (Me)
o Remove double REPORT_LUNS from cpqfcTSstructs.h (Me)
o Fix up potential oops in udp short packet logging (Me)
| doesn't affect mainline.
o sysrq updates. (James Simmons)
o Replace fbcon-cfb with software accels. (James Simmons)
o Allow ACPI configuration on x86-64 (Me)
o Drop the radix pagecache stuff.
(Was an ancient version, and a pita to maintain)
o nfs3 compile fix. (Al Viro)
o pnpbios updates. (Thomas Hood)


2.5.5-dj3
o Merge 2.5.6pre2
o Merge 2.4.19pre2
| Drop various arch bits. (See above)
| Drop LVM bits.
| Drop some SIS bits that rejected -- James please take a look.
| Fix some obvious sillies as per 2.4.19pre2ac1
o Add missing struct initialiser to nls_cp850 (OGAWA Hirofumi)
o natsemi rx/tx ring buffer confusion fixup. (Jeff Garzik)
o Backout bogus isofs makefile changes. (Me)
o Fix sys_shmdt return code. (Andreas Schwab)
o Numerous net driver last_rx = jiffies fixes. (Me, Celso Gonzalez)
o msync/mprotect POSIX fixes. (Thorsten Kukuk)
o Small SCSI generic updates. (Douglas Gilbert)
o Allow short commands in SCSI debug driver. (Douglas Gilbert)
o Update helptext for Machine Check. (Paul Gortmaker)
o Updated email address. (Thomas Molina)
o Fix potential oops in swapfile code. (Andries Brouwer)
o Reset/Reservation handling fixes. (James Bottomley)
o Various console fixes. (James Simmons)
o Update numerous fb drivers to new api. (James Simmons)
o Twin inquiry mode for SCSI LUN scanning. (Patrick Mansfield,
Douglas Gilbert)
o Update DMI / local APIC fixes. (Mikael Pettersson)
o Emu10k1 OSS highmem dma fixes. (Daniel Bertrand)
o Export dparent_lock (Petr Vandrovec)
o Clean up struct page shrinkage. (Rik van Riel, Dave Miller)
o More Config.help updates. (Steven Cole)
o Make highmem pte a boolean. (Steven Cole)
o opl3sa2 modular compile fix. (Zwane Mwaikambo)
o es18xx compile fix. (Zwane Mwaikambo)
o nfsd modular compile fix. (Stelian Pop)
o Use list heads for task list. (Brian Gerst)
o Fix up various compilation warnings. (Roberto Nibali)
o Power management for 3c509. (Zwane Mwaikambo)
o Small kbuild cleanup. (John Levon)


--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs


2002-03-17 18:43:22

by Douglas Gilbert

[permalink] [raw]
Subject: Re: Linux 2.5.6-dj1

Dave wrote:
> This one hasn't had any testing beyond "it compiles"

Compiled here but didn't link (SMP) :-(
page_cache_release undefined multiple times in mm/mm.o


There are over 30 scsi subsystem patches backed up in
your tree. Some are over 2 months old. Could
some (or perhaps all) of them get promoted to the
main tree? It is very difficult to write further patches
against the scsi subsystem with this sort of backlog.

Other subsystems (e.g. usb and ieee1394) might appreciate
some of those changes since they asked for them.

Doug Gilbert

------------------------------------------
patch-2.5.7-pre2 shows _one_ file patched in the
scsi subsystem:
patching file drivers/scsi/ide-scsi.c

patch-2.5.6-dj1 has the following list. Some of them
are superficial (e.g. "s/then/than/") but most have
substance:
patching file drivers/scsi/3w-xxxx.c
patching file drivers/scsi/3w-xxxx.h
patching file drivers/scsi/53c7,8xx.c
patching file drivers/scsi/53c7xx.c
patching file drivers/scsi/Config.help
patching file drivers/scsi/Config.in
patching file drivers/scsi/Makefile
patching file drivers/scsi/README.st
patching file drivers/scsi/aacraid/Makefile
patching file drivers/scsi/aacraid/README
patching file drivers/scsi/aacraid/TODO
patching file drivers/scsi/aacraid/aachba.c
patching file drivers/scsi/aacraid/aacraid.h
patching file drivers/scsi/aacraid/commctrl.c
patching file drivers/scsi/aacraid/comminit.c
patching file drivers/scsi/aacraid/commsup.c
patching file drivers/scsi/aacraid/dpcsup.c
patching file drivers/scsi/aacraid/linit.c
patching file drivers/scsi/aacraid/rx.c
patching file drivers/scsi/aacraid/sap1sup.c
patching file drivers/scsi/advansys.c
patching file drivers/scsi/aha1542.c
patching file drivers/scsi/aic7xxx/aic7xxx_linux.c
patching file drivers/scsi/aic7xxx/aicasm/aicdb.h
patching file drivers/scsi/aic7xxx_old.c
patching file drivers/scsi/cpqfcTSstructs.h
patching file drivers/scsi/dtc.c
patching file drivers/scsi/eata.c
patching file drivers/scsi/fdomain.c
patching file drivers/scsi/g_NCR5380.c
patching file drivers/scsi/g_NCR5380.h
patching file drivers/scsi/i60uscsi.c
patching file drivers/scsi/ibmmca.c
patching file drivers/scsi/ide-scsi.c
patching file drivers/scsi/imm.c
patching file drivers/scsi/ips.c
patching file drivers/scsi/megaraid.c
patching file drivers/scsi/mesh.c
patching file drivers/scsi/osst.c
patching file drivers/scsi/osst.h
patching file drivers/scsi/pas16.c
patching file drivers/scsi/pci2000.c
patching file drivers/scsi/pcmcia/aha152x_stub.c
patching file drivers/scsi/pcmcia/fdomain_stub.c
patching file drivers/scsi/pcmcia/qlogic_stub.c
patching file drivers/scsi/ppa.c
patching file drivers/scsi/scsi.c
patching file drivers/scsi/scsi.h
patching file drivers/scsi/scsi_debug.c
patching file drivers/scsi/scsi_debug.h
patching file drivers/scsi/scsi_error.c
patching file drivers/scsi/scsi_merge.c
patching file drivers/scsi/scsi_scan.c
patching file drivers/scsi/sd.c
patching file drivers/scsi/sg.c
patching file drivers/scsi/sgiwd93.c
patching file drivers/scsi/sr.c
patching file drivers/scsi/st.c
patching file drivers/scsi/sun3_scsi.c
patching file drivers/scsi/sym53c8xx_2/sym_hipd.c
patching file drivers/scsi/sym53c8xx_defs.h
patching file drivers/scsi/t128.c
patching file drivers/scsi/tmscsim.c
patching file drivers/scsi/u14-34f.c
patching file drivers/scsi/wd33c93.c

2002-03-17 20:43:26

by Dave Jones

[permalink] [raw]
Subject: Re: Linux 2.5.6-dj1

On Sun, Mar 17, 2002 at 01:44:16PM -0500, Douglas Gilbert wrote:

> Compiled here but didn't link (SMP) :-(
> page_cache_release undefined multiple times in mm/mm.o

Probably a side-effect of me removing the radix tree patch.
I'll look into this.

> There are over 30 scsi subsystem patches backed up in
> your tree. Some are over 2 months old. Could
> some (or perhaps all) of them get promoted to the
> main tree?

Indeed. Once Linus returns from vacation, I'll be doing a
patch-bombing on a larger scale than usual 8-)

Any bits I'm uncertain of, I'll bounce your way first for
clarification, deal ?

--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs

2002-03-17 23:59:05

by James Bottomley

[permalink] [raw]
Subject: Re: Linux 2.5.6-dj1

Douglas Gilbert <[email protected]> said:
> There are over 30 scsi subsystem patches backed up in
> your tree. Some are over 2 months old. Could
> some (or perhaps all) of them get promoted to the
> main tree?

[email protected] said:
> Indeed. Once Linus returns from vacation, I'll be doing a
> patch-bombing on a larger scale than usual 8-)
> Any bits I'm uncertain of, I'll bounce your way first for
> clarification, deal ?

This really brings up a very important point. As Doug has said before, since
there are quite a few patches floating around on the SCSI list and other
discussion of structural alteration in there, could we not have our own
bitkeeper repository for the SCSI subsystem to keep them all in (and so that
we can roll our own patch sets from)?

Jens, you're now listed as the SCSI maintainer so you are the person who is
supposed to serve as a conduit to Linus for all of the SCSI patches. You
would also be the logical person to look after a SCSI bitkeeper repository, so
will you look after one for us? I can volunteer to help, and I'm sure others
will.

I'm getting pretty tired of the snide comments from HW manufacturers and other
OS interests about the state of our SCSI subsystem (not to mention
unfavourable comparisons from the IDE guys). Perhaps something like this
could galvanise the community and serve as a catalyst for a general
streamlining of the subsystem.

James


2002-03-18 04:21:16

by Douglas Gilbert

[permalink] [raw]
Subject: Re: Linux 2.5.6-dj1

Dave Jones wrote:
>
> On Sun, Mar 17, 2002 at 01:44:16PM -0500, Douglas Gilbert wrote:
>
> > Compiled here but didn't link (SMP) :-(
> > page_cache_release undefined multiple times in mm/mm.o
>
> Probably a side-effect of me removing the radix tree patch.
> I'll look into this.

Dave,
Sorry, false alarm; "make mrproper" fixed that problem.

So here are some results from testing on my SMP box
(dual Celeron abit); all these worked:

- advansys driver (*) holds my root fs
- aha1542 driver (*) lightly tested
- imm (*) 100 MB parallel port zip drive
- scsi reset [part of James's reservation+reset patch]
- usb-storage to an ATA maxtor 6L040J2 disk via a lava
external enclosure
- usb-storage to a sandisk (8MB compact flash card)
- usb-storage to a casio digital camera
- scsi_debug driver

Notes:
- those drivers marked with (*) are broken on SMP
machines in 2.5.7-pre2
- unfortunately my ieee1394 card is another box,
the sbp2 driver works in 2.5.7-pre2 (although
it does have some quirks: my lava box claims
that an ATA disk is SCSI 6 compliant!)
- as can be seen, several other subsystems depend on
the scsi subsystem. My guess is the scsi subsystem
will be seeing more ATA disks (and cd/dvd writers)
in external enclosures via usb-2 and 1394 in the
future
- obviously the scsi mid level patches (I'm thinking of
the report_lun+twin_inquiry) are not causing any
problems.
- I noticed the sym53c8xx-2 driver has been updated.
There is a dc-390u3w on another machine here. If
there is any problem, I'll contact you

So all in all the scsi subsystem looks pretty good in
"dj1". I did lose some time when my keyboard + mouse
disappeared (.config problems). For those still having
trouble finding their mouse read Documentation/input/input.txt

> > There are over 30 scsi subsystem patches backed up in
> > your tree. Some are over 2 months old. Could
> > some (or perhaps all) of them get promoted to the
> > main tree?
>
> Indeed. Once Linus returns from vacation, I'll be doing a
> patch-bombing on a larger scale than usual 8-)
>
> Any bits I'm uncertain of, I'll bounce your way first for
> clarification, deal ?

Fine.

Doug Gilbert