Linux 2.4.20-pre1

Linux 2.4.20-pre1

[Marcelo Tosatti]

So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
stuff.

2.4.20 will be a much faster release cycle than 2.4.19 was.




Summary of changes from v2.4.19 to v2.4.20-pre1
============================================

<jejb@mulgrave.(none)> (02/03/16 1.159.1.1)
	53c700

<achirica@ttd.net> (02/05/31 1.445.1.13)
	airo wireless net driver update:

<rob@osinvestor.com> (02/06/05 1.537.4.1)
	Sparc32 code cleanups.

<rob@osinvestor.com> (02/06/05 1.537.4.2)
	Sparc32 code cleanups.

<davem@nuts.ninka.net> (02/06/07 1.537.4.3)
	Sparc: Fix copy_{to,from}_user return value handling.

<davem@nuts.ninka.net> (02/06/07 1.537.4.4)
	Sparc64: readv/writev SuS compliance fix for sparc32 compat.

<baccala@vger.freesoft.org> (02/06/07 1.537.4.5)
	DBRI driver: Add T7903 doc URL.

<kanojsarcar@yahoo.com> (02/06/07 1.537.4.6)
	Sparc64: Fix module symbols when stack debugging is on.

<jejb@mulgrave.(none)> (02/06/09 1.159.1.2)
	[SCSI 53c700] bux fix in tag starvation, cosmetic cleanup of set_depth

<jejb@mulgrave.(none)> (02/06/09 1.159.1.3)
	[SCSI 53c700] update version to 2.8

<ak@muc.de> (02/06/12 1.537.1.8)
	RTNETLINK: Allow non-root to receive.

<davem@nuts.ninka.net> (02/06/12 1.537.1.9)
	MAINTAINERS: Remove Andi from networking as per his request.

<rusty@rustcorp.com.au> (02/06/12 1.537.1.10)
	ipv4/route.c: Cleanup ip_rt_acct_read

<jgarzik@rum.normnet.org> (02/06/14 1.537.7.1)
	net driver 8139cp updates:

<wstinson@infonie.fr> (02/06/14 1.537.9.1)
	[janitor] update the isicom.c multiport serial driver to

<wstinson@infonie.fr> (02/06/14 1.537.7.3)
	[janitor] update the ray_cs.c PCMCIA client driver for the Raylink wireless LAN card

<wstinson@infonie.fr> (02/06/14 1.537.7.4)
	[janitor] update the ni65 network driver to

<wstinson@infonie.fr> (02/06/14 1.537.7.5)
	[janitor] update the sdlamain Multiprotocol WAN Link Driver to

<wstinson@infonie.fr> (02/06/14 1.537.9.2)
	[janitor] update the DAC960 Driver for Mylex DAC960/AcceleRAID/eXtremeRAID PCI RAID Controllers

<wstinson@infonie.fr> (02/06/14 1.537.7.6)
	[janitor] update the eexpress.c net driver to

<wstinson@infonie.fr> (02/06/14 1.537.7.7)
	[janitor] update the comx-hw-comx wan driver to remove call to check_region and check the status of call to

<wstinson@infonie.fr> (02/06/14 1.537.7.8)
	[janitor] update the eepro Intel EtherExpress Pro/10 device driver to

<wstinson@infonie.fr> (02/06/14 1.537.7.9)
	[janitor] update the atarilance Ethernet driver for VME Lance cards on the Atari to check the result of request_irq and exit in case of error.

<wstinson@infonie.fr> (02/06/14 1.537.7.10)
	[janitor] update the yam hamradio driver to

<jgarzik@rum.normnet.org> (02/06/14 1.537.10.1)
	Support dumping NIC-specific stats in 8139cp net driver

<jes@wildopensource.com> (02/06/15 1.537.1.11)
	Tigon3: Use unsigned type for dest_idx_unmasked in tg3_recycle_rx.

<jes@wildopensource.com> (02/06/15 1.537.1.12)
	Tigon3: MAX_WAIT_CNT is too large.

<edward_peng@dlink.com.tw> (02/06/15 1.537.7.12)
	dl2k gige net driver updates:

<michaelw@foldr.org> (02/06/17 1.537.4.7)
	sparc64: Use SUNW,power-off to power off some Ultra systems.

<ctindel@cup.hp.com> (02/06/18 1.537.1.13)
	drivers/net/bonding.c: Check ethtool then mii ioctl to determine link status.

<davem@nuts.ninka.net> (02/06/18 1.537.1.14)
	NET: Backport 2.5.x NAPI infrastructure to 2.4.x

<davem@nuts.ninka.net> (02/06/19 1.537.1.15)
	Tigon3: On 32-bit just wrap low 32bits of stats if we overflow.

<jgarzik@mandrakesoft.com> (02/06/20 1.537.12.3)
	Add register dumping and NIC-specific stats to 8139too net driver

<jgarzik@mandrakesoft.com> (02/06/20 1.537.12.4)
	Fix 8139too net driver register dump

<wa@almesberger.net> (02/06/21 1.537.1.16)
	include/net/dsfield.h: Remove dead code.

<jmorris@intercode.com.au> (02/06/25 1.537.1.17)
	NETLINK: Add unicast release notifier.

<davem@nuts.ninka.net> (02/06/25 1.537.1.18)
	SunHME: Register IRQ with netdev->name as string.

<davem@nuts.ninka.net> (02/07/11 1.537.1.19)
	Add netif_receive_skb-like interface for VLAN hw accel.

<davem@nuts.ninka.net> (02/07/11 1.537.1.20)
	Tigon3: Add NAPI support.

<kiran@in.ibm.com> (02/07/12 1.537.1.21)
	net/core/dst.c: dst_total only needs to exist if RT_CACHE_DEBUG >= 2

<rml@tech9.net> (02/07/15 1.537.1.22)
	net/socket.c: Kill memory leak in sock_fasync

<uzi@uzix.org> (02/07/16 1.537.4.8)
	SunHME: Make module license visible when not-PCI.

<davem@nuts.ninka.net> (02/07/16 1.537.4.9)
	arch/sparc64/defconfig: Update.

<robert.olsson@data.slu.se> (02/07/18 1.537.1.23)
	PKTGEN: Updates to version 1.2, work mostly from Ben Greear.

<davem@nuts.ninka.net> (02/07/18 1.537.1.24)
	PKTGEN: Use htonl instead of __constant_htonl.

<kuznet@ms2.inr.ac.ru> (02/07/18 1.537.1.25)
	PKT SCHED: Add HTB scheduler by Martin Devera.

<ecd@skynet.be> (02/07/18 1.537.4.10)
	SPARC64: Fix bugs in ioctl32 registration.

<szepe@pinerecords.com> (02/07/18 1.537.4.11)
	SPARC: Dynamically size SRMMU nocache page pool.

<ebrower@resilience.com> (02/07/19 1.537.1.26)
	SK98LIN: Fix oops in procfs handling if no cards probed.

<rob@osinvestor.com> (02/07/19 1.537.4.12)
	SPARC: Minor header file cleanups.

<rob@osinvestor.com> (02/07/19 1.537.4.13)
	floppy.h: Remove unused empty virtual_dma_init.

<robert.olsson@data.slu.se> (02/07/19 1.537.1.27)
	PKTGEN: Update documentation.

<davem@nuts.ninka.net> (02/07/19 1.537.1.28)
	TIGON3: Finish up NAPI implementation.

<davem@nuts.ninka.net> (02/07/19 1.537.1.29)
	PKTGEN: u64 is not necessarily a long long

<davem@nuts.ninka.net> (02/07/19 1.537.1.30)
	HTB PKTSCHED: u64 is not necessarily a long long

<davem@nuts.ninka.net> (02/07/19 1.537.1.31)
	PKTGEN: Fix need_resched for 2.4.x, more u64 printf fixes.

<davem@nuts.ninka.net> (02/07/19 1.537.1.32)
	PKTGEN: More need_resched 2.4.x fixes

<davem@nuts.ninka.net> (02/07/22 1.537.1.33)
	net/ipv4/route.c: Handle large offsets properly in procfs read operation.

<davem@nuts.ninka.net> (02/07/22 1.537.4.14)
	drivers/sbus/char/openprom.c: Verify user len in copyin_string.

<rob@osinvestor.com> (02/07/22 1.537.4.15)
	arch/sparc/config.in: Remove commented out LVM bits.

<davem@nuts.ninka.net> (02/07/30 1.537.4.16)
	Openprom: Cast nagative tests properly.

<davem@nuts.ninka.net> (02/07/30 1.537.4.17)
	OpenPROM: Cast to ssize_t not int.

<davem@nuts.ninka.net> (02/07/30 1.537.4.18)
	OpenPROM: Kill len check, it is pointless.

<davem@nuts.ninka.net> (02/07/31 1.537.4.19)
	OpenPROM: Sigh, put the length overflow check back it is needed.

<rusty@rustcorp.com.au> (02/08/02 1.582.2.84)
	[PATCH] wan/sdla_chdlc.c oops fix

<mauelshagen@sistina.com> (02/08/02 1.582.2.85)
	[PATCH] LVM 1.0.5 driver update for 2.4.19-rc3

<hch@lst.de> (02/08/02 1.582.2.86)
	[PATCH] remove unused label in fs/dnotify.c

<rusty@rustcorp.com.au> (02/08/02 1.582.2.87)
	[PATCH] ip_nat_core.c - fix compiler warning

<rusty@rustcorp.com.au> (02/08/02 1.582.2.88)
	[PATCH] 3c509.c - 1_2

<rusty@rustcorp.com.au> (02/08/02 1.582.2.89)
	[PATCH] 2.4 i_size_high fixup

<rusty@rustcorp.com.au> (02/08/02 1.582.2.90)
	[PATCH] remove agpgart_be.c unused variables

<rusty@rustcorp.com.au> (02/08/02 1.582.2.91)
	[PATCH] namespace.c - compiler warning

<hch@infradead.org> (02/08/02 1.582.2.92)
	[PATCH] small inline assembly fix for gcc 3.1 (ffs)

<szepe@pinerecords.com> (02/08/02 1.582.2.93)
	[PATCH] reserve nocache based on RAM size

<flo@rfc822.org> (02/08/02 1.582.2.94)
	[PATCH] 2.4.19-rc5 cyclades.c one liner

<garloff@suse.de> (02/08/02 1.582.2.95)
	[PATCH] IDE: memset kmalloced gendisk structures

<agrover@groveronline.com> (02/08/02 1.582.2.96)
	[PATCH] Put Intel cache-detection descriptors in a table

<hch@infradead.org> (02/08/02 1.582.2.97)
	[PATCH] proper boot-time messages for P4 Xeon

<bunk@fs.tum.de> (02/08/02 1.582.2.98)
	[PATCH] document that cmd64x.c supports the CMD649 and CMD680

<jo-lkml@suckfuell.net> (02/08/02 1.592)
	[PATCH] Correct locking on IO stats accounting

<willy@w.ods.org> (02/08/03 1.594)
	[PATCH] Reorder TOSHIBA TC35815 in Config.in

<rob@osinvestor.com> (02/08/03 1.595)
	[PATCH] watchdog flags

<kaos@ocs.com.au> (02/08/03 1.596)
	[PATCH] 2.4.19 include/linux/vmalloc.h for highmem

<marcel@holtmann.org> (02/08/03 1.596.2.1)
	[PATCH] Bluetooth Subsystem PC Card drivers update

<felipewd@terra.com.br> (02/08/03 1.596.3.1)
	Add Wake-On-Lan support to 8139cp net driver

<jgarzik@mandrakesoft.com> (02/08/03 1.596.3.2)
	Temporarily disable MTU-change-while-active support in 8139cp

<felipewd@terra.com.br> (02/08/03 1.596.3.3)
	Update 8139cp net driver to enable legacy Rx/TX command register

<felipewd@terra.com.br> (02/08/03 1.596.3.4)
	Add suspend/resume support to 8139cp net driver

<davidm@hpl.hp.com> (02/08/03 1.596.3.5)
	Update eepro100 net drvr to enable rx DMA without causing

<jgarzik@mandrakesoft.com> (02/08/03 1.596.3.6)
	Add Intel e100 net driver.

<jgarzik@mandrakesoft.com> (02/08/03 1.596.3.7)
	Add e1000 gige net driver.

<jgarzik@mandrakesoft.com> (02/08/03 1.596.3.8)
	Move e100 net driver after eepro100, in kernel image link order.

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.601)
	[PATCH] PATCH: 2.4.20-pre1 - parisc specific include files

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.602)
	[PATCH] PATCH: 2.4.20pre1 - parisc gsc bus

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.603)
	[PATCH] PATCH: 2.4.20-pre1 HP HIL drivers

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.604)
	[PATCH] PATCH: Update the MPT fusion drivers to the vendors latest

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.605)
	[PATCH] PATCH: 2.4/2.5 - Fix endianness on 3c503

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.606)
	[PATCH] PATCH: remove unneeded parisc magic in acenic

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.607)
	[PATCH] PATCH: depca update

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.608)
	[PATCH] PATCH: 2.4/2.5 - Fix traps on alpha when using ewrk3

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.609)
	[PATCH] PATCH: fix gcc warnings in baycom_epp

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.610)
	[PATCH] PATCH: 2.4/2.5 Fix endianness in hp-plus

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.611)
	[PATCH] PATCH: fix multiple gcc 3.1 warnings in irda

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.612)
	[PATCH] PATCH: 2.4/2.5 fix endianness on smc boards

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.613)
	[PATCH] PATCH: 2.4/2.5 fix endianness in wd.c

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.614)
	[PATCH] PATCH: 2.4 - move nubus to static inline

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.615)
	[PATCH] PATCH: 2.4 update gsc parallel port - from maintainer

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.616)
	[PATCH] PATCH: update the aacraid driver

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.617)
	[PATCH] PATCH: fix aic7xxx build when PCI=n

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.618)
	[PATCH] PATCH: eata scsi - update from author

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.619)
	[PATCH] PATCH: fix typo in ncr53c8xx

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.620)
	[PATCH] PATCH: u14-34f scsi driver update from author

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.621)
	[PATCH] PATCH: driver for harmony audio

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.622)
	[PATCH] PATCH: trident audio updates

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.623)
	[PATCH] PATCH: fix undefined C in ixj driver

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.624)
	[PATCH] PATCH: fix gcc 3.1 warnings

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.625)
	[PATCH] PATCH: update sti frame buffer

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.626)
	[PATCH] PATCH: pull the rest of sti into a subdirectory and update it

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.627)
	[PATCH] PATCH: Beos file systems (befs)

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.628)
	[PATCH] PATCH: SOM binary load (parisc specific)

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.629)
	[PATCH] PATCH: Add EFI partition support

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.630)
	[PATCH] PATCH: more hp specific include files

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.631)
	[PATCH] PATCH: Fix gcc 3.1 warnings in vlan

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.632)
	[PATCH] PATCH: fix formatting

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.633)
	[PATCH] PATCH: gcc 3.1 warning fixes for generic_serial

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.634)
	[PATCH] PATCH: HP specific drivers

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.635)
	[PATCH] PATCH: fix compiler warning in mxser

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.636)
	[PATCH] PATCH: (resend for 2.4.20-pre1 as asked) synclink pcmcia

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.637)
	[PATCH] PATCH: PDC (parisc bios) console driver

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.638)
	[PATCH] PATCH: fix compiler warnings

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.639)
	[PATCH] PATCH: fix stallion failing to load

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.640)
	[PATCH] PATCH: fix sx driver compile warnings

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.641)
	[PATCH] PATCH: fix standards compliance bugs in the tty layer

<alan@irongate.swansea.linux.org.uk> (02/08/05 1.642)
	[PATCH] PATCH: clean up umem further

<marcelo@plucky.distro.conectiva> (02/08/05 1.643)
	Changed makefile to .20-pre1

<marcelo@plucky.distro.conectiva> (02/08/05 1.644)
	Import patch revert-acenic-change.patch

<ak@muc.de> (02/08/05 1.645)
	[PATCH] AMD 8111 PCI IDS

<ak@muc.de> (02/08/05 1.646)
	[PATCH] AMD 8111 sound support

<ak@muc.de> (02/08/05 1.647)
	[PATCH] x86-64 auto_fs support

<ak@muc.de> (02/08/05 1.648)
	[PATCH] bluetooth flags warning fixes

<ak@muc.de> (02/08/05 1.649)
	[PATCH] cciss interrupt flags warnings

<ak@muc.de> (02/08/05 1.650)
	[PATCH] CONFIG_ISA for drivers/char

<ak@muc.de> (02/08/05 1.651)
	[PATCH] CONFIG_ISA for IDE

<ak@muc.de> (02/08/05 1.652)
	[PATCH] CONFIG_ISA for radio drivers

<ak@muc.de> (02/08/05 1.653)
	[PATCH] Configuration fixes for net drivers

<ak@muc.de> (02/08/05 1.654)
	[PATCH] DRM 64bit fixes

<ak@muc.de> (02/08/05 1.655)
	[PATCH] Early console support for x86-64

<ak@muc.de> (02/08/05 1.656)
	[PATCH] x86-64 ELF format name

<ak@muc.de> (02/08/05 1.657)
	[PATCH] Ftape 64bit/x86-64 fixes

<ak@muc.de> (02/08/05 1.658)
	[PATCH] CONFIG_ISA for i386

<ak@muc.de> (02/08/05 1.659)
	[PATCH] Interrupt flags fixes for IEEE1394

<ak@muc.de> (02/08/05 1.660)
	[PATCH] x86-64 support in ipc/

<ak@muc.de> (02/08/05 1.661)
	[PATCH] 64bit fixes for drivers/isdn

<ak@muc.de> (02/08/05 1.662)
	[PATCH] 64bit fixes for the megaraid driver

<ak@muc.de> (02/08/05 1.663)
	[PATCH] paride interrupt flags fixes

<ak@muc.de> (02/08/05 1.664)
	[PATCH] 64bit warning fixes for PCI

<ak@muc.de> (02/08/05 1.665)
	[PATCH] synclink interrupt flag fixes

<ak@muc.de> (02/08/05 1.666)
	[PATCH] 64bit fixes for drivers/video

<ak@muc.de> (02/08/05 1.667)
	[PATCH] vsyscall/HPET support for x86-64

<ak@muc.de> (02/08/05 1.668)
	[PATCH] 64bit WAN driver fixes

<ak@muc.de> (02/08/05 1.669)
	[PATCH] wavelan 64bit warning

<ak@muc.de> (02/08/05 1.670)
	[PATCH] x86-64 core changes

<pkot@linuxnews.pl> (02/08/05 1.671)
	[PATCH] remove the warning from include/linux/dcache.h

<ak@muc.de> (02/08/05 1.673)
	[PATCH] 64bit dpt driver fixes

<elenstev@mesatop.com> (02/08/05 1.674)
	[PATCH] 2.4.19,

<maxk@qualcomm.com> (02/08/05 1.675)
	[PATCH] core fixes

<maxk@qualcomm.com> (02/08/05 1.676)
	[PATCH] L2CAP fixes

<maxk@qualcomm.com> (02/08/05 1.677)
	[PATCH] SCO fixes

<maxk@qualcomm.com> (02/08/05 1.678)
	[PATCH] HCI USB driver update

<maxk@qualcomm.com> (02/08/05 1.679)
	[PATCH] BNEP support

<Kai.Makisara@kolumbus.fi> (02/08/05 1.680)
	[PATCH] SCSI tape patch for 2.4.19

<marcelo@plucky.distro.conectiva> (02/08/05 1.681)
	Remove NAPI for now

Re: Linux 2.4.20-pre1

[Ben Greear]

Marcelo Tosatti wrote:
> So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
> stuff.
> 
> 2.4.20 will be a much faster release cycle than 2.4.19 was.

Two questions:  I see change logs about NAPI going in, and then
NAPI being removed.  I assume it is removed...but maybe it will
be back soon?

Second:  Where is the patch?  I looked on kernel.org and didn't
find it.  If it's going to be there shortly, that's fine, I'll
keep checking back.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>       <Ben_Greear AT excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

Re: Linux 2.4.20-pre1

[Marcelo Tosatti]

On Mon, 5 Aug 2002, Ben Greear wrote:

> Marcelo Tosatti wrote:
> > So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
> > stuff.
> >
> > 2.4.20 will be a much faster release cycle than 2.4.19 was.
>
> Two questions:  I see change logs about NAPI going in, and then
> NAPI being removed.  I assume it is removed...but maybe it will
> be back soon?

I want arguments from Davem to include NAPI. Changing the drivers is a
reason for me to _not_ want it in.

But lets see if Davem can convince me ;)

> Second:  Where is the patch?  I looked on kernel.org and didn't
> find it.  If it's going to be there shortly, that's fine, I'll
> keep checking back.

Maybe at davem's CVS repo?

Re: Linux 2.4.20-pre1

[David S. Miller]

   From: Marcelo Tosatti <marcelo@conectiva.com.br>
   Date: Tue, 6 Aug 2002 08:32:59 -0300 (BRT)
   
   > Second:  Where is the patch?  I looked on kernel.org and didn't
   > find it.  If it's going to be there shortly, that's fine, I'll
   > keep checking back.
   
   Maybe at davem's CVS repo?

I don't use CVS anymore...

Re: Linux 2.4.20-pre1

[Jason Lunz]

Will you consider using one of those perl scripts to condense the
changelogs a little bit?

Jason

Re: Linux 2.4.20-pre1

[CaT]

On Tue, Aug 06, 2002 at 10:58:03AM -0400, Jason Lunz wrote:
> Will you consider using one of those perl scripts to condense the
> changelogs a little bit?

Please use them. -Please-...

-- 
| You wake up. The room is spinning very gently | >i
| round your head. Or at least it would be if   | You have:
| you could see it which you can't.             |   a splitting headache
|                                               |   no tea
| It is pitch black                             |

Re: Linux 2.4.20-pre1

[Ben Greear]

Marcelo Tosatti wrote:
> 
> On Mon, 5 Aug 2002, Ben Greear wrote:
> 
> 
>>Marcelo Tosatti wrote:
>>
>>>So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
>>>stuff.
>>>
>>>2.4.20 will be a much faster release cycle than 2.4.19 was.
>>
>>Two questions:  I see change logs about NAPI going in, and then
>>NAPI being removed.  I assume it is removed...but maybe it will
>>be back soon?
> 
> 
> I want arguments from Davem to include NAPI. Changing the drivers is a
> reason for me to _not_ want it in.
> 
> But lets see if Davem can convince me ;)

Well, I hope he does, and I hope it really works :)

The patch I was looking for was the pre-patch you put out.
I see it's on kernel.org now, so no problem there.

Ben
> 
> 
>>Second:  Where is the patch?  I looked on kernel.org and didn't
>>find it.  If it's going to be there shortly, that's fine, I'll
>>keep checking back.
> 
> 
> Maybe at davem's CVS repo?
> 


-- 
Ben Greear <greearb@candelatech.com>       <Ben_Greear AT excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

NAPI (was Re: Linux 2.4.20-pre1)

[Jeff Garzik]

Marcelo Tosatti wrote:
> I want arguments from Davem to include NAPI. Changing the drivers is a
> reason for me to _not_ want it in.
> 
> But lets see if Davem can convince me ;)



8139too needs it for flood protection.  I also have a patch for sundance 
which fixes the issue with the quad port and implements RX polling for 
flood protection.

Basically, NAPI --should not-- affect any system that is not using a 
NAPI driver.  Think of NAPI as a net driver libary -- if your driver 
doesn't use it, you don't know it's there at all.  And currently tg3 is 
the only 2.4 driver using NAPI.

NAPI saves people manually implementing polling in each driver, for 
flood and DoS protection that is needed in 2.4.  The flood ceiling is 
much lower without NAPI, due to having the CPU overhead of interrupt 
handlers.  NAPI also eliminates the need for lots of code to support all 
sorts of NIC hardware interrupt mitigation variants.

	Jeff

Re: Linux 2.4.20-pre1

[Adrian Bunk]

The changes to drivers/net/tg3.c in -pre1 broke the compilation:

<--  snip  -->

...
gcc -D__KERNEL__ -I/home/bunk/linux/kernel-2.4/linux-2.4.19-full/include
-Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing
-fno-common -pipe -mpreferred-stack-boundary=2 -march=k6   -nostdinc -I
/usr/lib/gcc-lib/i386-linux/2.95.4/include -DKBUILD_BASENAME=tg3  -c -o
tg3.o tg3.c
In file included from tg3.c:25:
/home/bunk/linux/kernel-2.4/linux-2.4.19-full/include/linux/if_vlan.h: In
function `__vlan_hwaccel_rx':
/home/bunk/linux/kernel-2.4/linux-2.4.19-full/include/linux/if_vlan.h:186:
warning: implicit declaration of function `netif_receive_skb'
tg3.c: In function `tg3_poll':
tg3.c:1936: structure has no member named `quota'
tg3.c:1937: structure has no member named `quota'
tg3.c:1942: structure has no member named `quota'
tg3.c:1949: warning: implicit declaration of function `netif_rx_complete'
tg3.c: In function `tg3_interrupt_main_work':
tg3.c:1976: warning: implicit declaration of function
`netif_rx_schedule_prep'
tg3.c:1979: warning: implicit declaration of function
`__netif_rx_schedule'
tg3.c: In function `tg3_init_one':
tg3.c:5991: structure has no member named `poll'
tg3.c:5992: structure has no member named `weight'
make[3]: *** [tg3.o] Error 1
make[3]: Leaving directory `/home/bunk/linux/kernel-2.4/linux-2.4.19-full/drivers/net'

<--  snip  -->

cu
Adrian

-- 

You only think this is a free country. Like the US the UK spends a lot of
time explaining its a free country because its a police state.
								Alan Cox

Re: Linux 2.4.20-pre1

[J.A. Magallon]

[-- Attachment #1: Type: text/plain, Size: 662 bytes --]


On 2002.08.06 Marcelo Tosatti wrote:
>
>So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
>stuff.
>
>
><jgarzik@mandrakesoft.com> (02/08/03 1.596.3.6)
>	Add Intel e100 net driver.
>
><jgarzik@mandrakesoft.com> (02/08/03 1.596.3.7)
>	Add e1000 gige net driver.
>

Configure.help missing for those. Attached.

btw, I don't like this also, but is the official one...

-- 
J.A. Magallon             \   Software is like sex: It's better when it's free
mailto:jamagallon@able.es  \                    -- Linus Torvalds, FSF T-shirt
Linux werewolf 2.4.19-jam0, Mandrake Linux 9.0 (Cooker) for i586
gcc (GCC) 3.2 (Mandrake Linux 9.0 3.2-0.2mdk)

[-- Attachment #2: intel-eth-help.diff.gz --]
[-- Type: application/x-gzip, Size: 1475 bytes --]

Re: Linux 2.4.20-pre1

[J.A. Magallon]

On 2002.08.06 Marcelo Tosatti wrote:
>
>So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
>stuff.
>

Something is missing in the network merge:

werewolf:/usr/src/linux# grep -r netif_receive_skb *
drivers/net/tg3.c:                      netif_receive_skb(skb);
Binary file drivers/net/e1000/e1000_main.o matches
Binary file drivers/net/e1000/e1000.o matches
include/linux/if_vlan.h:        //return (polling ? netif_receive_skb(skb) : netif_rx(skb));

So I had to:

--- linux/include/linux/if_vlan.h.orig	2002-08-07 02:57:36.000000000 +0200
+++ linux/include/linux/if_vlan.h	2002-08-07 02:58:07.000000000 +0200
@@ -183,7 +183,8 @@
 		break;
 	};
 
-	return (polling ? netif_receive_skb(skb) : netif_rx(skb));
+	//return (polling ? netif_receive_skb(skb) : netif_rx(skb));
+	return netif_rx(skb);
 }
 
 static inline int vlan_hwaccel_rx(struct sk_buff *skb,

To make e1000 build. But tg3 uses n_r_skb directly, so it is not useful for
that.

-- 
J.A. Magallon             \   Software is like sex: It's better when it's free
mailto:jamagallon@able.es  \                    -- Linus Torvalds, FSF T-shirt
Linux werewolf 2.4.19-jam0, Mandrake Linux 9.0 (Cooker) for i586
gcc (GCC) 3.2 (Mandrake Linux 9.0 3.2-0.2mdk)

Re: Linux 2.4.20-pre1

[Bryan Whitehead]

[-- Attachment #1: Type: text/plain, Size: 435 bytes --]

Marcelo Tosatti wrote:
> So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
> stuff.
> 
> 2.4.20 will be a much faster release cycle than 2.4.19 was.
> 
> 

This didn't make 2.4.19. Just a spelling error. I tried the maintainer 
but got no reply...

(yea yea... i just like clean logs... ) :)

;)

-- 
Bryan Whitehead
SysAdmin - JPL - Interferometry Systems and Technology
Phone: 818 354 2903
driver@jpl.nasa.gov

[-- Attachment #2: natsemi.patch --]
[-- Type: text/plain, Size: 400 bytes --]

--- linux/drivers/net/natsemi.c.orig	Tue Aug  6 18:50:13 2002
+++ linux/drivers/net/natsemi.c	Tue Aug  6 18:50:38 2002
@@ -1685,7 +1685,7 @@
 			np->tx_config += 2;
 		if (netif_msg_tx_err(np))
 			printk(KERN_NOTICE 
-				"%s: increased Tx theshold, txcfg %#08x.\n",
+				"%s: increased Tx threshold, txcfg %#08x.\n",
 				dev->name, np->tx_config);
 		writel(np->tx_config, ioaddr + TxConfig);
 	}

Re: Linux 2.4.20-pre1

[Tim Hockin]

> This didn't make 2.4.19. Just a spelling error. I tried the maintainer 
> but got no reply...
> 
> (yea yea... i just like clean logs... ) :)
> 
> ;)

ok, ok, It's going into my bk tree...sheesh :)

Re: Linux 2.4.20-pre1

[David S. Miller]

   Date: Tue, 6 Aug 2002 23:05:44 +0200 (CEST)
   From: Adrian Bunk <bunk@fs.tum.de>

   The changes to drivers/net/tg3.c in -pre1 broke the compilation:
 ...
   /home/bunk/linux/kernel-2.4/linux-2.4.19-full/include/linux/if_vlan.h:186:
   warning: implicit declaration of function `netif_receive_skb'
   tg3.c: In function `tg3_poll':

Because Marcelo took the tg3 NAPI changes but not the NAPI
infrastructure. :-)

I'm hoping since he's decided to take the generic NAPI changes
this compile failure will go away in -pre2

Re: NAPI (was Re: Linux 2.4.20-pre1)

[Marcus Sundberg]

Jeff Garzik <jgarzik@mandrakesoft.com> writes:

> I also have a patch for sundance which fixes the issue with
> the quad port

Which issue are you referring to?
The DFE-580-doesn't-work-without-USE_IO_OPS-defined, or the
drops-packets-even-under-moderate-load? If the latter I would
be very interrested in testing that code.

//Marcus
-- 
---------------------------------------+--------------------------
  Marcus Sundberg <marcus@ingate.com>  | Firewalls with SIP & NAT
 Firewall Developer, Ingate Systems AB |  http://www.ingate.com/

Re: Linux 2.4.20-pre1

[Erik Andersen]

On Mon Aug 05, 2002 at 07:40:56PM -0300, Marcelo Tosatti wrote:
> 
> So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
> stuff.
[------------snip------------]
> <alan@irongate.swansea.linux.org.uk> (02/08/05 1.629)
> 	[PATCH] PATCH: Add EFI partition support

Needs this to compile....

--- linux/include/asm-ia64/efi.h.orig	Sun Aug 11 01:41:10 2002
+++ linux/include/asm-ia64/efi.h	Sun Aug 11 01:43:38 2002
@@ -166,6 +166,9 @@
  *  EFI Configuration Table and GUID definitions
  */
 
+#define NULL_GUID    \
+    ((efi_guid_t) { 0x00000000, 0x0000, 0x0000, { 0x00, 0x00, 0x0, 0x00, 0x00, 0x00, 0x00, 0x00 }})
+
 #define MPS_TABLE_GUID    \
     ((efi_guid_t) { 0xeb9d2d2f, 0x2d88, 0x11d3, { 0x9a, 0x16, 0x0, 0x90, 0x27, 0x3f, 0xc1, 0x4d }})
 

> <rusty@rustcorp.com.au> (02/08/02 1.582.2.91)
> 	[PATCH] namespace.c - compiler warning

This patch is wrong....

--- linux/fs/namespace.c.orig	Sun Aug 11 01:50:52 2002
+++ linux/fs/namespace.c	Sun Aug 11 01:51:04 2002
@@ -29,8 +29,6 @@
 static int hash_mask, hash_bits;
 static kmem_cache_t *mnt_cache; 
 
-extern void init_rootfs(void);
-
 static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry)
 {
 	unsigned long tmp = ((unsigned long) mnt / L1_CACHE_BYTES);
--- linux/include/linux/namespace.h.orig	Sat Aug  3 17:14:31 2002
+++ linux/include/linux/namespace.h	Sat Aug  3 18:46:43 2002
@@ -38,5 +38,6 @@
 	atomic_inc(&namespace->count);
 }
 
+int __init init_rootfs(void);
 #endif
 #endif


 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: Linux 2.4.20-pre1

[Christoph Hellwig]

On Sun, Aug 11, 2002 at 02:57:17AM -0600, Erik Andersen wrote:
> > 	[PATCH] namespace.c - compiler warning
> 
> This patch is wrong....

Why?

Re: Linux 2.4.20-pre1

[Erik Andersen]

On Sun Aug 11, 2002 at 10:16:17AM +0100, Christoph Hellwig wrote:
> On Sun, Aug 11, 2002 at 02:57:17AM -0600, Erik Andersen wrote:
> > > 	[PATCH] namespace.c - compiler warning
> > 
> > This patch is wrong....
> 
> Why?

Because it adds function prototype that doesn't match the
actual function....  

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: Linux 2.4.20-pre1

[Erik Andersen]

On Sun Aug 11, 2002 at 10:16:17AM +0100, Christoph Hellwig wrote:
> On Sun, Aug 11, 2002 at 02:57:17AM -0600, Erik Andersen wrote:
> > > 	[PATCH] namespace.c - compiler warning
> > 
> > This patch is wrong....
> 
> Why?

I guess it would be more clear if I had said that namespace.c
in 2.4.20-pre1 was broken my the above mentioned patch, and
my patch fixes the problem...

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: Linux 2.4.20-pre1

[Erik Andersen]

On Mon Aug 05, 2002 at 07:40:56PM -0300, Marcelo Tosatti wrote:
> <achirica@ttd.net> (02/05/31 1.445.1.13)
> 	airo wireless net driver update:

Doesn't this make more sense?

--- linux/drivers/net/wireless/airo.c.orig	Sun Aug 11 03:59:28 2002
+++ linux/drivers/net/wireless/airo.c	Sun Aug 11 03:59:46 2002
@@ -191,12 +191,6 @@
 #ifndef RUN_AT
 #define RUN_AT(x) (jiffies+(x))
 #endif
-#ifndef PDE
-static inline struct proc_dir_entry *PDE(const struct inode *inode)
-{
-	return inode->u.generic_ip;
-}
-#endif
 
 
 /* These variables are for insmod, since it seems that the rates

diff -urN linux-2.4.19.orig/include/linux/proc_fs.h linux-2.4.19/include/linux/proc_fs.h
--- linux-2.4.19.orig/include/linux/proc_fs.h	Fri Aug  2 18:39:45 2002
+++ linux-2.4.19/include/linux/proc_fs.h	Sun Aug 11 00:23:55 2002
@@ -209,4 +209,9 @@
 
 #endif /* CONFIG_PROC_FS */
 
+static inline struct proc_dir_entry *PDE(const struct inode *inode)
+{
+	return (struct proc_dir_entry *)inode->u.generic_ip;
+}
+

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: Linux 2.4.20-pre1

[Erik Andersen]

On Sun Aug 11, 2002 at 08:46:19PM +0100, Alan Cox wrote:
> On Sun, 2002-08-11 at 09:57, Erik Andersen wrote:
> > On Mon Aug 05, 2002 at 07:40:56PM -0300, Marcelo Tosatti wrote:
> > > 
> > > So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
> > > stuff.
> > [------------snip------------]
> > > <alan@irongate.swansea.linux.org.uk> (02/08/05 1.629)
> > > 	[PATCH] PATCH: Add EFI partition support
> > 
> > Needs this to compile....
> > 
> > --- linux/include/asm-ia64/efi.h.orig	Sun Aug 11 01:41:10 2002
> > +++ linux/include/asm-ia64/efi.h	Sun Aug 11 01:43:38 2002
> > @@ -166,6 +166,9 @@
> >   *  EFI Configuration Table and GUID definitions
> >   */
> >  
> > +#define NULL_GUID    \
> > +    ((efi_guid_t) { 0x00000000, 0x0000, 0x0000, { 0x00, 0x00, 0x0, 0x00, 0x00, 0x00, 0x00, 0x00 }})
> > +
> 
> Not a good plan. EFI can be used on non ia64 so NULL_GUID belongs
> somewhere else


I know.  But if you look at the source, the generic stuff is
directly including asm-ia64/efi.h, so this makes it compile.  I
agree having the generic stuff include asm-ia64 code is really
stupid, but changing that seemed far more invasive and frankly, I
don't even know what an "EFI" is, so I didn't want to mess with
it beyond making it compile per how it is currently doing things,

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: Linux 2.4.20-pre1

[Alan Cox]

On Sun, 2002-08-11 at 09:57, Erik Andersen wrote:
> On Mon Aug 05, 2002 at 07:40:56PM -0300, Marcelo Tosatti wrote:
> > 
> > So here goes -pre1, with a big -ac and x86-64 merges, plus other smaller
> > stuff.
> [------------snip------------]
> > <alan@irongate.swansea.linux.org.uk> (02/08/05 1.629)
> > 	[PATCH] PATCH: Add EFI partition support
> 
> Needs this to compile....
> 
> --- linux/include/asm-ia64/efi.h.orig	Sun Aug 11 01:41:10 2002
> +++ linux/include/asm-ia64/efi.h	Sun Aug 11 01:43:38 2002
> @@ -166,6 +166,9 @@
>   *  EFI Configuration Table and GUID definitions
>   */
>  
> +#define NULL_GUID    \
> +    ((efi_guid_t) { 0x00000000, 0x0000, 0x0000, { 0x00, 0x00, 0x0, 0x00, 0x00, 0x00, 0x00, 0x00 }})
> +

Not a good plan. EFI can be used on non ia64 so NULL_GUID belongs
somewhere else

[PATCH] radeonfb update vs 2.4.20-pre1

[Erik Andersen]

Here is an update to the radeonfb.  This is based on a patch 
from Peter Horton that was posted to the lkml back in April
    http://www.uwsg.indiana.edu/hypermail/linux/kernel/0204.1/0364.html

I have been carrying this patch along in my own tree since April
and I find it a huge improvement.  Using this patch, I find that
the radeonfb is quite fast, and colors are correct.

I have updated the patch to fit into 2.4.20-pre1, fixed a few
obvious little things, and reworked the mtrr handling so it
matches the behavior of the other framebuffers.  Any chance 
we could get this into 2.4.20?

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--


--- drivers/video/radeonfb.c.orig	Sun Aug 11 03:43:12 2002
+++ drivers/video/radeonfb.c	Sun Aug 11 03:43:18 2002
@@ -22,6 +22,13 @@
  *
  *	Special thanks to ATI DevRel team for their hardware donations.
  *
+ * 	2002-04-02	Added MTRR support. Fixed 8bpp acceleration. Added
+ * 			acceleration for 16/32bpp. Applied fix from XFree86
+ * 			for hard crash on accelerator reset. Fixed up the
+ * 			colour stuff. Peter Horton <pdh@colonel-panic.org>
+ * 	2002-04-10	Make ypan work. More colour fixes, all modes >8bpp
+ * 			now DIRECTCOLOR. Match up CRTC and accelerator
+ * 			pitch.
  */
 
 
@@ -73,18 +80,24 @@
 #include <video/fbcon.h> 
 #include <video/fbcon-cfb8.h>
 #include <video/fbcon-cfb16.h>
-#include <video/fbcon-cfb24.h>
 #include <video/fbcon-cfb32.h>
+#ifdef CONFIG_MTRR
+#include <asm/mtrr.h>
+#endif
 
 #include "radeon.h"
 
 
-#define DEBUG	0
+#define FIX_DEPTH_15		1
+
+#define DEBUG			0
+
+#define _RTRACE(f,a...)		do{printk("radeonfb: " f "\n", ##a);}while(0)
 
 #if DEBUG
-#define RTRACE		printk
+#define RTRACE(f,a...)		_RTRACE(f,##a)
 #else
-#define RTRACE		if(0) printk
+#define RTRACE(f,a...)
 #endif
 
 
@@ -299,17 +312,16 @@
 
 	struct ram_info ram;
 
+#if 0
         u32 hack_crtc_ext_cntl;
         u32 hack_crtc_v_sync_strt_wid;
+#endif
 
 #if defined(FBCON_HAS_CFB16) || defined(FBCON_HAS_CFB32)
         union {
 #if defined(FBCON_HAS_CFB16)
                 u_int16_t cfb16[16];
 #endif
-#if defined(FBCON_HAS_CFB24)
-                u_int32_t cfb24[16];
-#endif  
 #if defined(FBCON_HAS_CFB32)
                 u_int32_t cfb32[16];
 #endif  
@@ -474,11 +486,10 @@
 }
 
 
-static inline int var_to_depth(const struct fb_var_screeninfo *var)
+static __inline__ int var_to_depth(const struct fb_var_screeninfo *var)
 {
-	if (var->bits_per_pixel != 16)
-		return var->bits_per_pixel;
-	return (var->green.length == 6) ? 16 : 15;
+	return (var->bits_per_pixel == 16 && var->green.length == 5)? 
+	    15 : var->bits_per_pixel;
 }
 
 
@@ -531,6 +542,7 @@
 #define radeon_engine_reset()		_radeon_engine_reset(rinfo)
 
 
+#if 0
 static __inline__ u8 radeon_get_post_div_bitval(int post_div)
 {
         switch (post_div) {
@@ -552,6 +564,7 @@
                         return 0x02;
         }
 }
+#endif
 
 
 
@@ -594,14 +607,25 @@
         
 static char fontname[40] __initdata;
 static char *mode_option __initdata;
-static char noaccel __initdata = 0;
+static char noaccel /*__initdata*/ = 0;
+static char fb16fix = 0;
 static int panel_yres __initdata = 0;
 static char force_dfp __initdata = 0;
 static struct radeonfb_info *board_list = NULL;
+#ifdef CONFIG_MTRR
+static int enable_mtrr = 1;
+static int mtrr_handle;
+#endif
 
 #ifdef FBCON_HAS_CFB8
 static struct display_switch fbcon_radeon8;
 #endif
+#ifdef FBCON_HAS_CFB16
+static struct display_switch fbcon_radeon16;
+#endif
+#ifdef FBCON_HAS_CFB32
+static struct display_switch fbcon_radeon32;
+#endif
 
 
 /*
@@ -731,6 +755,12 @@
 			force_dfp = 1;
 		} else if (!strncmp(this_opt, "panel_yres:", 11)) {
 			panel_yres = simple_strtoul((this_opt+11), NULL, 0);
+		} else if (!strncmp(this_opt, "fb16fix", 7)) {
+			fb16fix = 1;
+#ifdef CONFIG_MTRR
+		} else if (!strncmp(this_opt, "nomtrr", 6)) {
+			enable_mtrr = 0;
+#endif
                 } else
 			mode_option = this_opt;
         }
@@ -755,7 +785,7 @@
 	u32 tmp;
 	int i, j;
 
-	RTRACE("radeonfb_pci_register BEGIN\n");
+	RTRACE("radeonfb_pci_register BEGIN");
 
 	rinfo = kmalloc (sizeof (struct radeonfb_info), GFP_KERNEL);
 	if (!rinfo) {
@@ -915,7 +945,7 @@
 	rinfo->bios_seg = radeon_find_rom(rinfo);
 	radeon_get_pllinfo(rinfo, rinfo->bios_seg);
 
-	RTRACE("radeonfb: probed %s %dk videoram\n", (rinfo->ram_type), (rinfo->video_ram/1024));
+	RTRACE("probed %s %dk videoram", (rinfo->ram_type), (rinfo->video_ram/1024));
 
 #if !defined(__powerpc__)
 	radeon_get_moninfo(rinfo);
@@ -998,6 +1028,15 @@
 		return -ENODEV;
 	}
 
+#ifdef CONFIG_MTRR
+	if (enable_mtrr) {
+	    mtrr_handle = mtrr_add(rinfo->fb_base_phys, rinfo->video_ram, MTRR_TYPE_WRCOMB, 1);
+	    printk("radeonfb: MTRR enabled\n");
+	}
+#endif
+
+	/* XXX look at this re pitch :pdh */
+
 	if (!noaccel) {
 		/* initialize the engine */
 		radeon_engine_init (rinfo);
@@ -1029,7 +1068,7 @@
 			GET_MON_NAME(rinfo->crtDisp_type));
 	}
 
-	RTRACE("radeonfb_pci_register END\n");
+	RTRACE("radeonfb_pci_register END");
 
 	return 0;
 }
@@ -1046,6 +1085,12 @@
 	/* restore original state */
         radeon_write_mode (rinfo, &rinfo->init_state);
  
+#ifdef CONFIG_MTRR
+	if (enable_mtrr) {
+	    mtrr_del(mtrr_handle, rinfo->fb_base_phys, rinfo->video_ram);
+	    printk("radeonfb: MTRR disabled\n");
+	}
+#endif
         unregister_framebuffer ((struct fb_info *) rinfo);
                 
         iounmap ((void*)rinfo->mmio_base);
@@ -1157,7 +1202,7 @@
         	rinfo->pll.ppll_min = pll.PCLK_min_freq;
         	rinfo->pll.ppll_max = pll.PCLK_max_freq;
 
-		printk("radeonfb: ref_clk=%d, ref_div=%d, xclk=%d from BIOS\n",
+		_RTRACE("ref_clk=%d, ref_div=%d, xclk=%d from BIOS",
 			rinfo->pll.ref_clk, rinfo->pll.ref_div, rinfo->pll.xclk);
 	} else {
 #ifdef CONFIG_ALL_PPC
@@ -1177,7 +1222,7 @@
 			rinfo->pll.ppll_min = 12000;
 			rinfo->pll.ppll_max = 35000;
 
-			printk("radeonfb: ref_clk=%d, ref_div=%d, xclk=%d from OF\n",
+			_RTRACE("ref_clk=%d, ref_div=%d, xclk=%d from OF",
 				rinfo->pll.ref_clk, rinfo->pll.ref_div, rinfo->pll.xclk);
 
 			return;
@@ -1212,7 +1257,7 @@
 				break;
 		}
 
-		printk("radeonfb: ref_clk=%d, ref_div=%d, xclk=%d defaults\n",
+		_RTRACE("ref_clk=%d, ref_div=%d, xclk=%d defaults",
 			rinfo->pll.ref_clk, rinfo->pll.ref_div, rinfo->pll.xclk);
 	}
 }
@@ -1271,7 +1316,7 @@
 {
 #ifdef CONFIG_ALL_PPC
 	if (!radeon_get_EDID_OF(rinfo))
-		RTRACE("radeonfb: could not retrieve EDID from OF\n");
+		RTRACE("radeonfb: could not retrieve EDID from OF");
 #else
 	/* XXX use other methods later */
 #endif
@@ -1503,6 +1548,8 @@
 		rinfo->vSync_width = (unsigned short) ((tmp & FP_V_SYNC_WID_MASK) >>
 					FP_V_SYNC_WID_SHIFT);
 
+		/* XXX some more stuff needs to go here :pdh */
+
 		return 1;
 	}
 
@@ -1542,13 +1589,10 @@
 	radeon_fifo_wait (1);
 	OUTREG(DSTCACHE_MODE, 0);
 
-	/* XXX */
-	rinfo->pitch = ((rinfo->xres * (rinfo->bpp / 8) + 0x3f)) >> 6;
-
 	radeon_fifo_wait (1);
 	temp = INREG(DEFAULT_PITCH_OFFSET);
 	OUTREG(DEFAULT_PITCH_OFFSET, ((temp & 0xc0000000) | 
-				      (rinfo->pitch << 0x16)));
+				      (rinfo->pitch << 16)));
 
 	radeon_fifo_wait (1);
 	OUTREGP(DP_DATATYPE, 0, ~HOST_BIG_ENDIAN_EN);
@@ -1703,30 +1747,22 @@
         switch (disp->var.bits_per_pixel) {
 #ifdef FBCON_HAS_CFB8
                 case 8:
-                        disp->dispsw = &fbcon_cfb8;
+                        disp->dispsw = accel ? &fbcon_radeon8 : &fbcon_cfb8;
                         disp->visual = FB_VISUAL_PSEUDOCOLOR;
                         disp->line_length = disp->var.xres_virtual;
                         break;
 #endif
 #ifdef FBCON_HAS_CFB16
                 case 16:
-                        disp->dispsw = &fbcon_cfb16;
+                        disp->dispsw = accel ? &fbcon_radeon16 : &fbcon_cfb16;
                         disp->dispsw_data = &rinfo->con_cmap.cfb16;
                         disp->visual = FB_VISUAL_DIRECTCOLOR;
                         disp->line_length = disp->var.xres_virtual * 2;
                         break;
 #endif  
-#ifdef FBCON_HAS_CFB32       
-                case 24:
-                        disp->dispsw = &fbcon_cfb24;
-                        disp->dispsw_data = &rinfo->con_cmap.cfb24;
-                        disp->visual = FB_VISUAL_DIRECTCOLOR;
-                        disp->line_length = disp->var.xres_virtual * 4;
-                        break;
-#endif
 #ifdef FBCON_HAS_CFB32
                 case 32:
-                        disp->dispsw = &fbcon_cfb32;
+                        disp->dispsw = accel ? &fbcon_radeon32 : &fbcon_cfb32;
                         disp->dispsw_data = &rinfo->con_cmap.cfb32;
                         disp->visual = FB_VISUAL_DIRECTCOLOR;
                         disp->line_length = disp->var.xres_virtual * 4;
@@ -1736,8 +1772,6 @@
                         printk ("radeonfb: setting fbcon_dummy renderer\n");
                         disp->dispsw = &fbcon_dummy;
         }
-                
-        return;
 }
                         
 
@@ -1759,6 +1793,7 @@
 
 
 
+#if 0
 static int radeonfb_do_maximize(struct radeonfb_info *rinfo,
                                 struct fb_var_screeninfo *var,
                                 struct fb_var_screeninfo *v,
@@ -1819,9 +1854,18 @@
                                 
         return 0;
 }
+#endif
                         
 
 
+static void set_palette_entry(struct radeonfb_info *rinfo, unsigned idx, unsigned red, unsigned grn, unsigned blu)
+{
+	OUTREG(PALETTE_INDEX, idx);
+	OUTREG(PALETTE_DATA, (red << 16) | (grn << 8) | blu);
+
+	udelay(1);	/* is this necessary ? */
+}
+
 /*
  * fb ops
  */
@@ -1844,7 +1888,7 @@
         fix->type_aux = disp->type_aux;
         fix->visual = disp->visual;
 
-        fix->xpanstep = 8;
+        fix->xpanstep = 0;
         fix->ypanstep = 1;
         fix->ywrapstep = 0;
         
@@ -1880,7 +1924,7 @@
         struct radeonfb_info *rinfo = (struct radeonfb_info *) info;
         struct display *disp;
         struct fb_var_screeninfo v;
-        int nom, den, accel;
+        int accel, bytpp, pitch;
         unsigned chgvar = 0;
 
         disp = (con < 0) ? rinfo->info.disp : &fb_display[con];
@@ -1903,110 +1947,97 @@
         switch (v.bits_per_pixel) {
 		case 0 ... 8:
 			v.bits_per_pixel = 8;
-			break;
-		case 9 ... 16:
-			v.bits_per_pixel = 16;
-			break;
-		case 17 ... 24:
-			v.bits_per_pixel = 24;
-			break;
-		case 25 ... 32:
-			v.bits_per_pixel = 32;
-			break;
-		default:
-			return -EINVAL;
-	}
-
-	switch (var_to_depth(&v)) {
-#ifdef FBCON_HAS_CFB8
-                case 8:
-                        nom = den = 1;
-                        disp->line_length = v.xres_virtual;
-                        disp->visual = FB_VISUAL_PSEUDOCOLOR; 
                         v.red.offset = v.green.offset = v.blue.offset = 0;
                         v.red.length = v.green.length = v.blue.length = 8;
                         v.transp.offset = v.transp.length = 0;
-                        break;
-#endif
-                        
-#ifdef FBCON_HAS_CFB16
-		case 15:
-			nom = 2;
-			den = 1;
-			disp->line_length = v.xres_virtual * 2;
-			disp->visual = FB_VISUAL_DIRECTCOLOR;
-			v.red.offset = 10;
-			v.green.offset = 5;
-			v.red.offset = 0;
-			v.red.length = v.green.length = v.blue.length = 5;
-			v.transp.offset = v.transp.length = 0;
 			break;
-                case 16:
-                        nom = 2;
-                        den = 1;
-                        disp->line_length = v.xres_virtual * 2;
-                        disp->visual = FB_VISUAL_DIRECTCOLOR;
-                        v.red.offset = 11;
+		case 9 ... 16:
+#if FIX_DEPTH_15
+			if(v.bits_per_pixel < 16)
+				v.green.length = 5;
+#endif
+			if(v.green.length != 5 && v.green.length != 6)
+				v.green.length = 5;
+			v.bits_per_pixel = 16;
+                        v.red.offset = v.green.length + 5;
                         v.green.offset = 5;
                         v.blue.offset = 0;
                         v.red.length = 5;
-                        v.green.length = 6;
                         v.blue.length = 5;
                         v.transp.offset = v.transp.length = 0;
-                        break;  
-#endif
-                        
-#ifdef FBCON_HAS_CFB24
-                case 24:
-                        nom = 4;
-                        den = 1;
-                        disp->line_length = v.xres_virtual * 3;
-                        disp->visual = FB_VISUAL_DIRECTCOLOR;
+			break;
+		case 17 ... 32:
+			v.bits_per_pixel = 32;
                         v.red.offset = 16;
                         v.green.offset = 8;
                         v.blue.offset = 0;
                         v.red.length = v.blue.length = v.green.length = 8;
+                        /* v.transp.offset = 24; v.transp.length = 8; */
                         v.transp.offset = v.transp.length = 0;
-                        break;
-#endif
-#ifdef FBCON_HAS_CFB32
-                case 32:
-                        nom = 4;
-                        den = 1;
-                        disp->line_length = v.xres_virtual * 4;
-                        disp->visual = FB_VISUAL_DIRECTCOLOR;
-                        v.red.offset = 16;
-                        v.green.offset = 8;
-                        v.blue.offset = 0;
-                        v.red.length = v.blue.length = v.green.length = 8;
-                        v.transp.offset = 24;
-                        v.transp.length = 8;
-                        break;
-#endif
-                default:
-                        printk ("radeonfb: mode %dx%dx%d rejected, color depth invalid\n",
-                                var->xres, var->yres, var->bits_per_pixel);
-                        return -EINVAL;
-        }
+			break;
+		default:
+			return -EINVAL;
+	}
+
+        v.red.msb_right = v.green.msb_right = v.blue.msb_right = v.transp.msb_right = 0;
+
+	bytpp = v.bits_per_pixel >> 3;
+
+	if(v.xres > v.xres_virtual)
+		v.xres_virtual = v.xres;
+	if(v.yres > v.yres_virtual)
+		v.yres_virtual = v.yres;
+
+	/* XXX this is probably wrong for 24bpp, if we ever get it working :pdh */
+
+	pitch = (v.xres_virtual * bytpp + 63) & ~63;
+
+	if(pitch >= 8192)
+		return -EINVAL;
+
+	v.xres_virtual = pitch / bytpp;
+
+	if(v.xres_virtual < v.xres)
+		v.xres = v.xres_virtual;
+	
+	v.yres_virtual = rinfo->video_ram / pitch;
+
+	/* XXX there is some limit here, I chose 8192 :pdh */
 
+	if(v.yres_virtual >= 8192)
+		v.yres_virtual = 8191;
+
+	if(v.yres_virtual < v.yres)
+		v.yres = v.yres_virtual;
+
+#if 0
         if (radeonfb_do_maximize(rinfo, var, &v, nom, den) < 0)
                 return -EINVAL;  
+#endif
                 
+	if(v.xoffset)
+		return -EINVAL;
+
         if (v.xoffset < 0)
                 v.xoffset = 0;
         if (v.yoffset < 0)
                 v.yoffset = 0;
          
         if (v.xoffset > v.xres_virtual - v.xres)
-                v.xoffset = v.xres_virtual - v.xres - 1;
+		return -EINVAL;
                         
         if (v.yoffset > v.yres_virtual - v.yres)
-                v.yoffset = v.yres_virtual - v.yres - 1;
-         
-        v.red.msb_right = v.green.msb_right = v.blue.msb_right =
-                          v.transp.offset = v.transp.length =
-                          v.transp.msb_right = 0;
+		return -EINVAL;
                         
+	/* XXX this shouldn't be here :pdh */
+
+	disp->visual = (v.bits_per_pixel == 8 ? FB_VISUAL_PSEUDOCOLOR : FB_VISUAL_DIRECTCOLOR);
+	disp->line_length = v.xres_virtual * bytpp;
+	disp->ypanstep = 1;
+	disp->ywrapstep = 0;
+	disp->var.yres_virtual = v.yres_virtual;
+
+
         switch (v.activate & FB_ACTIVATE_MASK) {
                 case FB_ACTIVATE_TEST:
                         return 0;
@@ -2018,15 +2049,13 @@
         }
         
         memcpy (&disp->var, &v, sizeof (v));
-        
+
         if (chgvar) {     
-                radeon_set_dispsw(rinfo, disp);
 
-                if (noaccel)
-                        disp->scrollmode = SCROLL_YREDRAW;
-                else
-                        disp->scrollmode = 0;
-                
+		disp->scrollmode = accel ? 0 : SCROLL_YREDRAW;
+
+		radeon_set_dispsw(rinfo, disp);
+
                 if (info && info->changevar)
                         info->changevar(con);
         }
@@ -2093,6 +2122,8 @@
                                  struct fb_info *info)
 {
         struct radeonfb_info *rinfo = (struct radeonfb_info *) info;
+
+#if 0
         u32 offset, xoffset, yoffset;
                 
         xoffset = (var->xoffset + 7) & ~7;
@@ -2105,7 +2136,18 @@
         offset = ((yoffset * var->xres + xoffset) * var->bits_per_pixel) >> 6;
          
         OUTREG(CRTC_OFFSET, offset);
+#endif
+
+	/* XXX no X pan for the moment :pdh */
+
+	if(var->xoffset)
+		return -EINVAL;
         
+	if(var->yoffset + var->yres > var->yres_virtual)
+		return -EINVAL;
+
+	OUTREG(CRTC_OFFSET, var->yoffset * rinfo->pitch);
+
         return 0;
 }
 
@@ -2144,12 +2186,14 @@
                 do_install_cmap(con, info);
         }       
 
+#if 0
         /* XXX absurd hack for X to restore console */
         {   
 		OUTREGP(CRTC_EXT_CNTL, rinfo->hack_crtc_ext_cntl,
 			CRTC_HSYNC_DIS | CRTC_VSYNC_DIS | CRTC_DISPLAY_DIS);
                 OUTREG(CRTC_V_SYNC_STRT_WID, rinfo->hack_crtc_v_sync_strt_wid);
         }
+#endif
 
         return 0;
 }
@@ -2214,18 +2258,7 @@
 
 static int radeon_get_cmap_len (const struct fb_var_screeninfo *var)
 {
-        int rc = 256;            /* reasonable default */
-        
-        switch (var_to_depth(var)) {
-                case 15:
-                        rc = 32;
-                        break;
-		case 16:
-			rc = 64;
-			break;
-        }
-                
-        return rc;
+	return var->bits_per_pixel == 8 ? 256 : 16;
 }
 
 
@@ -2239,10 +2272,9 @@
 	if (regno > 255)
 		return 1;
      
- 	*red = (rinfo->palette[regno].red<<8) | rinfo->palette[regno].red; 
-    	*green = (rinfo->palette[regno].green<<8) | rinfo->palette[regno].green;
-    	*blue = (rinfo->palette[regno].blue<<8) | rinfo->palette[regno].blue;
-    	*transp = 0;
+ 	*red	= (unsigned) rinfo->palette[regno].red << 8;
+    	*green	= (unsigned) rinfo->palette[regno].green << 8;
+    	*blue	= (unsigned) rinfo->palette[regno].blue << 8;
 
 	return 0;
 }                            
@@ -2250,79 +2282,95 @@
 
 
 static int radeon_setcolreg (unsigned regno, unsigned red, unsigned green,
-                             unsigned blue, unsigned transp, struct fb_info *info)
+                             unsigned blue, unsigned alpha, struct fb_info *info)
 {
         struct radeonfb_info *rinfo = (struct radeonfb_info *) info;
-	u32 pindex;
 
-	if (regno > 255)
+	if(regno > 255)
 		return 1;
 
-	red >>= 8;
-	green >>= 8;
-	blue >>= 8;
-	rinfo->palette[regno].red = red;
-	rinfo->palette[regno].green = green;
-	rinfo->palette[regno].blue = blue;
+	red	>>= 8;
+	green	>>= 8;
+	blue	>>= 8;
+	alpha	>>= 8;
+
+	rinfo->palette[regno].red	= red;
+	rinfo->palette[regno].green	= green;
+	rinfo->palette[regno].blue	= blue;
+
+	/* ugly hack so that the software cursor works */
+
+	if(regno == 0) {
+		switch(rinfo->depth) {
+
+			case 16:
+				radeon_setcolreg(63, ~0, ~0, ~0, 0, info);
+				/* */
 
-        /* default */
-        pindex = regno;
-        
-	if (rinfo->bpp == 16) {
-		pindex = regno * 8;
+			case 15:
+				radeon_setcolreg(31, ~0, ~0, ~0, 0, info);
+				break;
 
-		if (rinfo->depth == 16 && regno > 63)
-			return 1;
-		if (rinfo->depth == 15 && regno > 31)
-			return 1;
+			case 32:
+				radeon_setcolreg(255, ~0, ~0, ~0, 0, info);
+				break;
+		}
+	}
 
-		/* For 565, the green component is mixed one order below */
-		if (rinfo->depth == 16) {
-	                OUTREG(PALETTE_INDEX, pindex>>1);
-       	         	OUTREG(PALETTE_DATA, (rinfo->palette[regno>>1].red << 16) |
-                        	(green << 8) | (rinfo->palette[regno>>1].blue));
-                	green = rinfo->palette[regno<<1].green;
-        	}
-	}
-
-	if (rinfo->depth != 16 || regno < 32) {
-		OUTREG(PALETTE_INDEX, pindex);
-		OUTREG(PALETTE_DATA, (red << 16) | (green << 8) | blue);
+	switch(rinfo->depth) {
+
+		case 15:
+			if(regno > 31)
+				return 1;
+			if(regno < 16)
+				rinfo->con_cmap.cfb16[regno] = (regno << 10) | (regno << 5) | regno;
+			regno <<= 3;
+			break;
+
+		case 16:
+
+			if(fb16fix) {
+
+				/* this is just another format of 15 bit mode, the LSB of the   *
+				 * green pixel data is ignored. this is required for "fbtv" and *
+				 * other apps to work correctly at 16bpp, but it means you land *
+				 * up with a sick looking boot logo. the real fix is to run at  *
+				 * 15bpp when you want to use those apps                   :pdh */
+
+				if(regno > 31)
+					return 1;
+				if(regno < 16)
+					rinfo->con_cmap.cfb16[regno] = (regno << 11) | (regno << 6) | regno;
+				regno <<= 3;
+				set_palette_entry(rinfo, regno | 4, red, green, blue);
+
+			} else {
+
+				if(regno > 63)
+					return 1;
+				if(regno < 16)
+					rinfo->con_cmap.cfb16[regno] = (regno << 11) | (regno << 5) | regno;
+				set_palette_entry(rinfo, regno << 2, rinfo->palette[regno >> 1].red,
+					green, rinfo->palette[regno >> 1].blue);
+				if(regno > 31)
+					return 0;
+				regno <<= 1;
+				green = rinfo->palette[regno].green;
+				regno <<= 2;
+			}
+			break;
+
+		case 32:
+			if(regno < 16)
+				rinfo->con_cmap.cfb32[regno] = (regno << 16) | (regno << 8) | regno;
+			break;
 	}
 
- 	if (regno < 16) {
-        	switch (rinfo->depth) {
-#ifdef FBCON_HAS_CFB16
-		        case 15:
-        			rinfo->con_cmap.cfb16[regno] = (regno << 10) | (regno << 5) |
-				                       	 	  regno;   
-			        break;
-		        case 16:
-        			rinfo->con_cmap.cfb16[regno] = (regno << 11) | (regno << 5) |
-				                       	 	  regno;   
-			        break;
-#endif
-#ifdef FBCON_HAS_CFB24
-                        case 24:
-                                rinfo->con_cmap.cfb24[regno] = (regno << 16) | (regno << 8) | regno;
-                                break;
-#endif
-#ifdef FBCON_HAS_CFB32
-	        	case 32: {
-            			u32 i;    
-   
-  		       		i = (regno << 8) | regno;
-            			rinfo->con_cmap.cfb32[regno] = (i << 16) | i;
-		        	break;
-        		}
-#endif
-		}
-        }
+	set_palette_entry(rinfo, regno, red, green, blue);
+
 	return 0;
 }
 
-
-
 static void radeon_save_state (struct radeonfb_info *rinfo,
                                struct radeon_regs *save)
 {
@@ -2373,8 +2421,10 @@
 	int primary_mon = PRIMARY_MONITOR(rinfo);
 	int depth = var_to_depth(mode);
 
+	bytpp = mode->bits_per_pixel >> 3;
 	rinfo->xres = mode->xres;
 	rinfo->yres = mode->yres;
+	rinfo->pitch = mode->xres_virtual * bytpp;
 	rinfo->pixclock = mode->pixclock;
 
 	hSyncStart = mode->xres + mode->right_margin;
@@ -2404,9 +2454,9 @@
 	h_sync_pol = sync & FB_SYNC_HOR_HIGH_ACT ? 0 : 1;
 	v_sync_pol = sync & FB_SYNC_VERT_HIGH_ACT ? 0 : 1;
 
-	RTRACE("hStart = %d, hEnd = %d, hTotal = %d\n",
+	RTRACE("hStart = %d, hEnd = %d, hTotal = %d",
 		hSyncStart, hSyncEnd, hTotal);
-	RTRACE("vStart = %d, vEnd = %d, vTotal = %d\n",
+	RTRACE("vStart = %d, vEnd = %d, vTotal = %d",
 		vSyncStart, vSyncEnd, vTotal);
 
 	hsync_wid = (hSyncEnd - hSyncStart) / 8;
@@ -2427,7 +2477,6 @@
 	cSync = mode->sync & FB_SYNC_COMP_HIGH_ACT ? (1 << 4) : 0;
 
 	format = radeon_get_dstbpp(depth);
-	bytpp = mode->bits_per_pixel >> 3;
 
 	if ((primary_mon == MT_DFP) || (primary_mon == MT_LCD))
 		hsync_fudge = hsync_fudge_fp[format-1];
@@ -2463,7 +2512,7 @@
 	newmode.crtc_v_sync_strt_wid = (((vSyncStart - 1) & 0xfff) |
 					 (vsync_wid << 16) | (v_sync_pol  << 23));
 
-	newmode.crtc_pitch = (mode->xres >> 3);
+	newmode.crtc_pitch = mode->xres_virtual >> 3;
 	newmode.crtc_pitch |= (newmode.crtc_pitch << 16);
 
 #if defined(__BIG_ENDIAN)
@@ -2479,12 +2528,9 @@
 	}
 #endif
 
-	rinfo->pitch = ((mode->xres * ((mode->bits_per_pixel + 1) / 8) + 0x3f)
-			& ~(0x3f)) / 64;
-
-	RTRACE("h_total_disp = 0x%x\t   hsync_strt_wid = 0x%x\n",
+	RTRACE("h_total_disp = 0x%x  hsync_strt_wid = 0x%x",
 		newmode.crtc_h_total_disp, newmode.crtc_h_sync_strt_wid);
-	RTRACE("v_total_disp = 0x%x\t   vsync_strt_wid = 0x%x\n",
+	RTRACE("v_total_disp = 0x%x  vsync_strt_wid = 0x%x",
 		newmode.crtc_v_total_disp, newmode.crtc_v_sync_strt_wid);
 
 	newmode.xres = mode->xres;
@@ -2493,8 +2539,10 @@
 	rinfo->bpp = mode->bits_per_pixel;
 	rinfo->depth = depth;
 
+#if 0
 	rinfo->hack_crtc_ext_cntl = newmode.crtc_ext_cntl;
 	rinfo->hack_crtc_v_sync_strt_wid = newmode.crtc_v_sync_strt_wid;
+#endif
 
 	if (freq > rinfo->pll.ppll_max)
 		freq = rinfo->pll.ppll_max;
@@ -2532,9 +2580,9 @@
 		newmode.ppll_div_3 = rinfo->fb_div | (post_div->bitvalue << 16);
 	}
 
-	RTRACE("post div = 0x%x\n", rinfo->post_div);
-	RTRACE("fb_div = 0x%x\n", rinfo->fb_div);
-	RTRACE("ppll_div_3 = 0x%x\n", newmode.ppll_div_3);
+	RTRACE("post div = 0x%x", rinfo->post_div);
+	RTRACE("fb_div = 0x%x", rinfo->fb_div);
+	RTRACE("ppll_div_3 = 0x%x", newmode.ppll_div_3);
 
 	/* DDA */
 	vclk_freq = round_div(rinfo->pll.ref_clk * rinfo->fb_div,
@@ -2554,8 +2602,8 @@
 	       xclk_per_trans) << (11 - useable_precision);
 	roff = xclk_per_trans_precise * (32 - 4);
 
-	RTRACE("ron = %d, roff = %d\n", ron, roff);
-	RTRACE("vclk_freq = %d, per = %d\n", vclk_freq, xclk_per_trans_precise);
+	RTRACE("ron = %d, roff = %d", ron, roff);
+	RTRACE("vclk_freq = %d, per = %d", vclk_freq, xclk_per_trans_precise);
 
 	if ((ron + rinfo->ram.rloop) >= roff) {
 		printk("radeonfb: error ron out of range\n");
@@ -2621,6 +2669,11 @@
 		newmode.tmds_crc = rinfo->init_state.tmds_crc;
 		newmode.tmds_transmitter_cntl = rinfo->init_state.tmds_transmitter_cntl;
 
+		/* XXX                                                  :pdh *
+		 * this doesn't work with some DFPs. a different driver that *
+		 * does work loads the FP CRTC registers from initial saved  *
+		 * values and doesn't tweak with tmds_transmitter_cntl ????? */
+
 		if (primary_mon == MT_LCD) {
 			newmode.lvds_gen_cntl |= (LVDS_ON | LVDS_BLON);
 			newmode.fp_gen_cntl &= ~(FP_FPON | FP_TMDS_EN);
@@ -2657,15 +2710,13 @@
 	int primary_mon = PRIMARY_MONITOR(rinfo);
 
 	/* blank screen */
-	OUTREGP(CRTC_EXT_CNTL, CRTC_DISPLAY_DIS | CRTC_VSYNC_DIS | CRTC_HSYNC_DIS,
-		~(CRTC_DISPLAY_DIS | CRTC_VSYNC_DIS | CRTC_HSYNC_DIS));
+	OUTREGP(CRTC_EXT_CNTL, mode->crtc_ext_cntl,
+		~(CRTC_HSYNC_DIS | CRTC_VSYNC_DIS | CRTC_DISPLAY_DIS));
 
 	for (i=0; i<9; i++)
 		OUTREG(common_regs[i].reg, common_regs[i].val);
 
 	OUTREG(CRTC_GEN_CNTL, mode->crtc_gen_cntl);
-	OUTREGP(CRTC_EXT_CNTL, mode->crtc_ext_cntl,
-		CRTC_HSYNC_DIS | CRTC_VSYNC_DIS | CRTC_DISPLAY_DIS);
 	OUTREGP(DAC_CNTL, mode->dac_cntl, DAC_RANGE_CNTL | DAC_BLANKING);
 	OUTREG(CRTC_H_TOTAL_DISP, mode->crtc_h_total_disp);
 	OUTREG(CRTC_H_SYNC_STRT_WID, mode->crtc_h_sync_strt_wid);
@@ -2743,7 +2794,7 @@
 	}
 
 	/* unblank screen */
-	OUTREG8(CRTC_EXT_CNTL + 1, 0);
+	OUTREG(CRTC_EXT_CNTL, mode->crtc_ext_cntl);
 
 	return;
 }
@@ -2966,6 +3017,8 @@
 	width *= fontwidth(p);
 	height *= fontheight(p);
 
+	dp_cntl = 0;
+
 	if (srcy < dsty) {
 		srcy += height - 1;
 		dsty += height - 1;
@@ -2992,22 +3045,8 @@
 }
 
 
-
-static void fbcon_radeon_clear(struct vc_data *conp, struct display *p,
-			       int srcy, int srcx, int height, int width)
+static void fbcon_radeon_clear(struct radeonfb_info *rinfo, int srcy, int srcx, int height, int width, u32 clr)
 {
-	struct radeonfb_info *rinfo = (struct radeonfb_info *)(p->fb_info);
-	u32 clr;
-
-	clr = attr_bgcol_ec(p, conp);
-	clr |= (clr << 8);
-	clr |= (clr << 16);
-
-	srcx *= fontwidth(p);
-	srcy *= fontheight(p);
-	width *= fontwidth(p);
-	height *= fontheight(p);
-
 	radeon_fifo_wait(6);
 	OUTREG(DP_GUI_MASTER_CNTL, (rinfo->dp_gui_master_cntl |
 				    GMC_BRUSH_SOLID_COLOR |
@@ -3021,12 +3060,24 @@
 }
 
 
-
 #ifdef FBCON_HAS_CFB8
+static void fbcon_radeon8_clear(struct vc_data *conp, struct display *p,
+		int srcy, int srcx, int height, int width)
+{
+	u32 clr;
+  
+	clr = attr_bgcol_ec(p, conp);
+	clr |= clr << 8;
+
+	fbcon_radeon_clear((struct radeonfb_info *) p->fb_info, srcy * fontheight(p), srcx * fontwidth(p),
+			height * fontheight(p), width * fontwidth(p), clr | (clr << 16));
+}
+
+
 static struct display_switch fbcon_radeon8 = {
 	setup:			fbcon_cfb8_setup,
 	bmove:			fbcon_radeon_bmove,
-	clear:			fbcon_radeon_clear,
+	clear:			fbcon_radeon8_clear,
 	putc:			fbcon_cfb8_putc,
 	putcs:			fbcon_cfb8_putcs,
 	revc:			fbcon_cfb8_revc,
@@ -3034,3 +3085,48 @@
 	fontwidthmask:		FONTWIDTH(4)|FONTWIDTH(8)|FONTWIDTH(12)|FONTWIDTH(16)
 };
 #endif
+
+#ifdef FBCON_HAS_CFB16
+static void fbcon_radeon16_clear(struct vc_data *conp, struct display *p,
+		int srcy, int srcx, int height, int width)
+{
+	u32 clr;
+
+	clr = ((u16 *) p->dispsw_data)[attr_bgcol_ec(p, conp)];
+
+	fbcon_radeon_clear((struct radeonfb_info *) p->fb_info, srcy * fontheight(p), srcx * fontwidth(p),
+			height * fontheight(p), width * fontwidth(p), clr | (clr << 16));
+}
+
+static struct display_switch fbcon_radeon16 = {
+	setup:			fbcon_cfb16_setup,
+	bmove:			fbcon_radeon_bmove,
+	clear:			fbcon_radeon16_clear,
+	putc:			fbcon_cfb16_putc,
+	putcs:			fbcon_cfb16_putcs,
+	revc:			fbcon_cfb16_revc,
+	clear_margins:		fbcon_cfb16_clear_margins,
+	fontwidthmask:		FONTWIDTH(4)|FONTWIDTH(8)|FONTWIDTH(12)|FONTWIDTH(16)
+};
+#endif
+
+#ifdef FBCON_HAS_CFB32
+static void fbcon_radeon32_clear(struct vc_data *conp, struct display *p,
+		int srcy, int srcx, int height, int width)
+{
+	fbcon_radeon_clear((struct radeonfb_info *) p->fb_info, srcy * fontheight(p), srcx * fontwidth(p),
+		height * fontheight(p), width * fontwidth(p), ((u32 *) p->dispsw_data)[attr_bgcol_ec(p, conp)]);
+}
+
+static struct display_switch fbcon_radeon32 = {
+	setup:			fbcon_cfb32_setup,
+	bmove:			fbcon_radeon_bmove,
+	clear:			fbcon_radeon32_clear,
+	putc:			fbcon_cfb32_putc,
+	putcs:			fbcon_cfb32_putcs,
+	revc:			fbcon_cfb32_revc,
+	clear_margins:		fbcon_cfb32_clear_margins,
+  	fontwidthmask:		FONTWIDTH(4)|FONTWIDTH(8)|FONTWIDTH(12)|FONTWIDTH(16)
+};
+#endif
+

Re: [PATCH] radeonfb update vs 2.4.20-pre1

[Ani Joshi]

These issues are addressed (or should be) in version 0.1.5 of radeonfb
which was posted to the fbdev list a month or so ago.  I am waiting on a
few other fixes I need to get done before I send an update to Marcelo, but
rest assured this will happen in the 2.4.20 cycle.


ani


On Sun, 11 Aug 2002, Erik Andersen wrote:

> Here is an update to the radeonfb.  This is based on a patch
> from Peter Horton that was posted to the lkml back in April
>     http://www.uwsg.indiana.edu/hypermail/linux/kernel/0204.1/0364.html
>
> I have been carrying this patch along in my own tree since April
> and I find it a huge improvement.  Using this patch, I find that
> the radeonfb is quite fast, and colors are correct.
>
> I have updated the patch to fit into 2.4.20-pre1, fixed a few
> obvious little things, and reworked the mtrr handling so it
> matches the behavior of the other framebuffers.  Any chance
> we could get this into 2.4.20?
>
>  -Erik
>
> --
> Erik B. Andersen             http://codepoet-consulting.com/
> --This message was written using 73% post-consumer electrons--
>

Re: [PATCH] radeonfb update vs 2.4.20-pre1

[Erik Andersen]

On Mon Aug 12, 2002 at 03:18:02AM -0700, Ani Joshi wrote:
> 
> These issues are addressed (or should be) in version 0.1.5 of radeonfb
> which was posted to the fbdev list a month or so ago.  I am waiting on a
> few other fixes I need to get done before I send an update to Marcelo, but
> rest assured this will happen in the 2.4.20 cycle.

Excellent thanks.  Is the latest and greatest posted somewhere?

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: Linux 2.4.20-pre1

[Paul P Komkoff Jr]

Replying to Alan Cox:
> Not a good plan. EFI can be used on non ia64 so NULL_GUID belongs
> somewhere else

maybe but even without NULL_GUID I still need to add asm-ia64 to
KBUILD_INCLUDE_DIRS while trying to kbuild-25ify -ac (and marcelo now)

hint-hint: maybe it IS worth moving to generic include ?

hmm

-- 
Paul P 'Stingray' Komkoff 'Greatest' Jr /// (icq)23200764 /// (http)stingr.net
  When you're invisible, the only one really watching you is you (my keychain)

Re: Linux 2.4.20-pre1

[Adrian Bunk]

On Mon, 5 Aug 2002, Marcelo Tosatti wrote:

>...
> Summary of changes from v2.4.19 to v2.4.20-pre1
> ============================================
>...
> <ak@muc.de> (02/08/05 1.657)
> 	[PATCH] Ftape 64bit/x86-64 fixes
>...

This broke the modular building of ftape. In -pre3:

<--  snip  -->

...
depmod: *** Unresolved symbols in
/lib/modules/2.4.20-pre3/kernel/drivers/char/ftape/lowlevel/ftape.o
depmod:         i8253_lock
...

<--  snip  -->


Alan made the following patch to fix it (already in -ac):


--- linux.20pre2/arch/i386/kernel/time.c	2002-08-13 13:58:33.000000000 +0100
+++ linux.20pre2-ac3/arch/i386/kernel/time.c	2002-08-12 15:17:04.000000000 +0100
@@ -31,6 +31,7 @@
  */

 #include <linux/errno.h>
+#include <linux/module.h>
 #include <linux/sched.h>
 #include <linux/kernel.h>
 #include <linux/param.h>
@@ -116,6 +117,8 @@

 spinlock_t i8253_lock = SPIN_LOCK_UNLOCKED;

+EXPORT_SYMBOL(i8253_lock);
+
 extern spinlock_t i8259A_lock;

 #ifndef CONFIG_X86_TSC



cu
Adrian