From: Andreas Schwab <hidden> Date: 2015-07-13 18:13:52
Finn Thain [off-list ref] writes:
BTW, I didn't change the SCSI ID location in NVRAM. This code says 16
whereas atari_scsi says 14. Which one is correct?
I think atari_scsi is wrong. The best source I could find
(http://www.gratifiant.com/nvram-falcon-t561185) places it after the
video mode byte, thus at byte 16.
Andreas.
--
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
"And now for something completely different."
Thanks for that. BTW, I googled a phrase from that page and found this one
http://toshyp.atari.org/en/004009.html
which may be closer to the source.
I'll send a patch if someone can offer to test such a change to atari_scsi
(or merely confirm that the SCSI ID setting in TOS does not match this_id
given in dmesg).
--
Hence "int" is 16-bits in all those references?
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
From: Michael Schmitz <schmitzmic@gmail.com> Date: 2015-07-22 03:52:17
Hi Finn,
I'm afraid I cannot test anything on Atari hardware at present - my
Falcon ate it's IDE disk partition table with all the fun that entails.
Haven't even begun to try and recover that yet.
If you send a patch I could build a kernel and send that to Christian
for testing (if he's got his Falcon up and running - might be a tad warm
in the attic for that, in fact).
Cheers,
Michael
Am 14.07.15 um 20:17 schrieb Finn Thain:
On Mon, 13 Jul 2015, Andreas Schwab wrote:
quoted
Finn Thain [off-list ref] writes:
quoted
BTW, I didn't change the SCSI ID location in NVRAM. This code says 16
whereas atari_scsi says 14. Which one is correct?
Thanks for that. BTW, I googled a phrase from that page and found this one
http://toshyp.atari.org/en/004009.html
which may be closer to the source.
I'll send a patch if someone can offer to test such a change to atari_scsi
(or merely confirm that the SCSI ID setting in TOS does not match this_id
given in dmesg).
From: Finn Thain <hidden> Date: 2015-07-22 04:22:46
On Wed, 22 Jul 2015, Michael Schmitz wrote:
Hi Finn,
I'm afraid I cannot test anything on Atari hardware at present - my
Falcon ate it's IDE disk partition table with all the fun that entails.
That doesn't sound good.
Haven't even begun to try and recover that yet.
If you send a patch I could build a kernel and send that to Christian
for testing (if he's got his Falcon up and running - might be a tad warm
in the attic for that, in fact).
Anyone with a suitable Atari, i.e. ATARIHW_PRESENT(TT_CLK), who can boot
both TOS and Linux could resolve the question. (Perhaps with an emulator?)
Any old kernel binary would do, since atari_scsi should print either
"HOSTID=n" or "this_id n" at startup.
If n doesn't agree with what TOS says about the host's SCSI ID, then I
think a trivial patch is safe enough. Especially if cat /proc/driver/nvram
produces a "SCSI host ID : m" that does agree with TOS.
Regards,
Finn
Cheers,
Michael
Am 14.07.15 um 20:17 schrieb Finn Thain:
quoted
On Mon, 13 Jul 2015, Andreas Schwab wrote:
quoted
Finn Thain [off-list ref] writes:
quoted
BTW, I didn't change the SCSI ID location in NVRAM. This code says 16
whereas atari_scsi says 14. Which one is correct?
Thanks for that. BTW, I googled a phrase from that page and found this one
http://toshyp.atari.org/en/004009.html
which may be closer to the source.
I'll send a patch if someone can offer to test such a change to atari_scsi
(or merely confirm that the SCSI ID setting in TOS does not match this_id
given in dmesg).
From: Christian T. Steigies <hidden> Date: 2015-07-22 14:33:44
On Wed, Jul 22, 2015 at 02:22:21PM +1000, Finn Thain wrote:
On Wed, 22 Jul 2015, Michael Schmitz wrote:
quoted
Hi Finn,
I'm afraid I cannot test anything on Atari hardware at present - my
Falcon ate it's IDE disk partition table with all the fun that entails.
That doesn't sound good.
quoted
Haven't even begun to try and recover that yet.
If you send a patch I could build a kernel and send that to Christian
for testing (if he's got his Falcon up and running - might be a tad warm
in the attic for that, in fact).
Anyone with a suitable Atari, i.e. ATARIHW_PRESENT(TT_CLK), who can boot
both TOS and Linux could resolve the question. (Perhaps with an emulator?)
The Falcon is not powered on currently but it should still work. What should
I test?
Any old kernel binary would do, since atari_scsi should print either
"HOSTID=n" or "this_id n" at startup.
If n doesn't agree with what TOS says about the host's SCSI ID, then I
think a trivial patch is safe enough. Especially if cat /proc/driver/nvram
produces a "SCSI host ID : m" that does agree with TOS.
From: Michael Schmitz <schmitzmic@gmail.com> Date: 2015-07-22 23:46:48
Hi Christian,
if I understand Finn right, he needs the SCSI host ID reported for the
Falcon SCSI chip, as found in this kernel log line:
scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq
15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { },
options { REAL_DMA SUPPORT_TAGS }
(see this_id; output from one of my ARAnyM instances, kernel 3.19rc6).
Also, the SCSI host ID reported by cat /proc/drivers/nvram (also 7, in
my case)
TOS reports the SCSI host ID during boot (when scannig for bootable
disks) IIRC. There may be some control panel to view the NVRAM
settings as well (stuff like default video mode was accessible from
the CT60 control panel so the SCSI host ID might be there as well).
Cheers,
Michael
On Thu, Jul 23, 2015 at 2:32 AM, Christian T. Steigies [off-list ref] wrote:
On Wed, Jul 22, 2015 at 02:22:21PM +1000, Finn Thain wrote:
quoted
On Wed, 22 Jul 2015, Michael Schmitz wrote:
quoted
Hi Finn,
I'm afraid I cannot test anything on Atari hardware at present - my
Falcon ate it's IDE disk partition table with all the fun that entails.
That doesn't sound good.
quoted
Haven't even begun to try and recover that yet.
If you send a patch I could build a kernel and send that to Christian
for testing (if he's got his Falcon up and running - might be a tad warm
in the attic for that, in fact).
Anyone with a suitable Atari, i.e. ATARIHW_PRESENT(TT_CLK), who can boot
both TOS and Linux could resolve the question. (Perhaps with an emulator?)
The Falcon is not powered on currently but it should still work. What should
I test?
quoted
Any old kernel binary would do, since atari_scsi should print either
"HOSTID=n" or "this_id n" at startup.
If n doesn't agree with what TOS says about the host's SCSI ID, then I
think a trivial patch is safe enough. Especially if cat /proc/driver/nvram
produces a "SCSI host ID : m" that does agree with TOS.
From: Finn Thain <hidden> Date: 2015-07-23 00:49:28
On Thu, 23 Jul 2015, Michael Schmitz wrote:
Hi Christian,
if I understand Finn right, he needs the SCSI host ID reported for the
Falcon SCSI chip, as found in this kernel log line:
What I'd really like is a regression test for this patch series. The SCSI
host ID issue is a separate one; these patches won't fix that bug.
scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq
15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { },
options { REAL_DMA SUPPORT_TAGS }
(see this_id; output from one of my ARAnyM instances, kernel 3.19rc6).
Also, the SCSI host ID reported by cat /proc/drivers/nvram (also 7, in
my case)
This seems inconclusive to me. Perhaps if you were to change a setting in
TOS such that the two values became unequal...
TOS reports the SCSI host ID during boot (when scannig for bootable
disks) IIRC. There may be some control panel to view the NVRAM settings
as well (stuff like default video mode was accessible from the CT60
control panel so the SCSI host ID might be there as well).
Can that be done in Aranym also?
--
Cheers,
Michael
On Thu, Jul 23, 2015 at 2:32 AM, Christian T. Steigies [off-list ref] wrote:
quoted
On Wed, Jul 22, 2015 at 02:22:21PM +1000, Finn Thain wrote:
quoted
On Wed, 22 Jul 2015, Michael Schmitz wrote:
quoted
Hi Finn,
I'm afraid I cannot test anything on Atari hardware at present - my
Falcon ate it's IDE disk partition table with all the fun that entails.
That doesn't sound good.
quoted
Haven't even begun to try and recover that yet.
If you send a patch I could build a kernel and send that to Christian
for testing (if he's got his Falcon up and running - might be a tad warm
in the attic for that, in fact).
Anyone with a suitable Atari, i.e. ATARIHW_PRESENT(TT_CLK), who can boot
both TOS and Linux could resolve the question. (Perhaps with an emulator?)
The Falcon is not powered on currently but it should still work. What should
I test?
quoted
Any old kernel binary would do, since atari_scsi should print either
"HOSTID=n" or "this_id n" at startup.
If n doesn't agree with what TOS says about the host's SCSI ID, then I
think a trivial patch is safe enough. Especially if cat /proc/driver/nvram
produces a "SCSI host ID : m" that does agree with TOS.
Christian
--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Christian T. Steigies <hidden> Date: 2015-07-23 09:44:32
On Wed, Jul 22, 2015 at 02:22:21PM +1000, Finn Thain wrote:
Anyone with a suitable Atari, i.e. ATARIHW_PRESENT(TT_CLK), who can boot
both TOS and Linux could resolve the question. (Perhaps with an emulator?)
Any old kernel binary would do, since atari_scsi should print either
"HOSTID=n" or "this_id n" at startup.
If n doesn't agree with what TOS says about the host's SCSI ID, then I
think a trivial patch is safe enough. Especially if cat /proc/driver/nvram
produces a "SCSI host ID : m" that does agree with TOS.
root@garkin:~>cat /proc/hardware
Model: Atari Falcon
System Memory: 522752K
510 MB at 0x01000000 (alternate RAM)
Detected hardware:
Falcon Shifter
Programmable Sound Generator
PCM 8 Bit Sound
CODEC Sound
SCSI Controller NCR5380 (Falcon style)
IDE Interface
8/16 Mhz Switch for FDC
Multi Function Peripheral MFP 68901
Serial Communications Controller SCC 8530
Paddle Interface
DMA Controller for SCC
Clock Chip MC146818A
Blitter
DSP56001 processor
root@garkin:~>dmesg |grep SCSI
[ 0.000000] Atari hardware found: VIDEL STDMA-SCSI ST_MFP YM2149 PCM CODEC DSP56K SCC_DMA SCC ANALOG_JOY BLITTER IDE TT_CLK FDC_SPEED
[ 0.410000] SCSI subsystem initialized
[ 0.850000] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 4.230000] Atari SCSI: resetting the SCSI bus... done
[ 6.750000] scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq 15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { }, options { REAL_DMA SUPPORT_TAGS }
root@garkin:~>cat /proc/driver/nvram
Checksum status : not valid
Boot preference : 0xff (undefined)
SCSI arbitration : on
SCSI host ID : 7
OS language : 255 (undefined)
Keyboard language: 255 (undefined)
Date format : 7 (undefined), 24h clock
Boot delay : 255s
Video mode : 4 colors, 40 columns, TV NTSC monitor
no overscan, compat. mode off
Let me know if you need more info.
Christian
From: Michael Schmitz <schmitzmic@gmail.com> Date: 2015-07-24 02:56:30
Hi Christian,
here's what Finn asked me to run as tests:
# dmesg | grep this_id > nvram.out
# cat /proc/driver/nvram >> nvram.out
# hexdump -C /dev/nvram >> nvram.out
# cp /dev/nvram /tmp/nvram
# cp /tmp/nvram /dev/nvram
# md5sum /dev/nvram /tmp/nvram >> nvram.out
What you sent so far looks OK. I've tested the change of SCSI ID
(using EmuTOS) along with a trivial patch to atari_scsi.c (replace
offset 14 by 16) and the driver now uses the stored ID properly. I'll
sent kernels to test Finn's NVRAM patch for regression ASAP.
Thanks for your offer of help!
Cheers,
Michael
On Thu, Jul 23, 2015 at 9:21 PM, Christian T. Steigies [off-list ref] wrote:
On Wed, Jul 22, 2015 at 02:22:21PM +1000, Finn Thain wrote:
quoted
Anyone with a suitable Atari, i.e. ATARIHW_PRESENT(TT_CLK), who can boot
both TOS and Linux could resolve the question. (Perhaps with an emulator?)
Any old kernel binary would do, since atari_scsi should print either
"HOSTID=n" or "this_id n" at startup.
If n doesn't agree with what TOS says about the host's SCSI ID, then I
think a trivial patch is safe enough. Especially if cat /proc/driver/nvram
produces a "SCSI host ID : m" that does agree with TOS.
root@garkin:~>cat /proc/hardware
Model: Atari Falcon
System Memory: 522752K
510 MB at 0x01000000 (alternate RAM)
Detected hardware:
Falcon Shifter
Programmable Sound Generator
PCM 8 Bit Sound
CODEC Sound
SCSI Controller NCR5380 (Falcon style)
IDE Interface
8/16 Mhz Switch for FDC
Multi Function Peripheral MFP 68901
Serial Communications Controller SCC 8530
Paddle Interface
DMA Controller for SCC
Clock Chip MC146818A
Blitter
DSP56001 processor
root@garkin:~>dmesg |grep SCSI
[ 0.000000] Atari hardware found: VIDEL STDMA-SCSI ST_MFP YM2149 PCM CODEC DSP56K SCC_DMA SCC ANALOG_JOY BLITTER IDE TT_CLK FDC_SPEED
[ 0.410000] SCSI subsystem initialized
[ 0.850000] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 4.230000] Atari SCSI: resetting the SCSI bus... done
[ 6.750000] scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq 15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { }, options { REAL_DMA SUPPORT_TAGS }
root@garkin:~>cat /proc/driver/nvram
Checksum status : not valid
Boot preference : 0xff (undefined)
SCSI arbitration : on
SCSI host ID : 7
OS language : 255 (undefined)
Keyboard language: 255 (undefined)
Date format : 7 (undefined), 24h clock
Boot delay : 255s
Video mode : 4 colors, 40 columns, TV NTSC monitor
no overscan, compat. mode off
Let me know if you need more info.
Christian
From: Finn Thain <hidden> Date: 2015-07-25 00:36:00
On Fri, 24 Jul 2015, Christian T. Steigies wrote:
Moin,
On Fri, Jul 24, 2015 at 02:56:26PM +1200, Michael Schmitz wrote:
quoted
here's what Finn asked me to run as tests:
# dmesg | grep this_id > nvram.out
# cat /proc/driver/nvram >> nvram.out
# hexdump -C /dev/nvram >> nvram.out
# cp /dev/nvram /tmp/nvram
# cp /tmp/nvram /dev/nvram
# md5sum /dev/nvram /tmp/nvram >> nvram.out
It seems I could successfully reset the nvram under TOS with bootconf.
Here is nvram.out, using the new kernel you sent me:
root@garkin:~>uname -a
Linux garkin 4.2.0-rc2-atari-269994-gc1e9d12 #327 Fri Jul 24 19:34:44 NZST 2015 m68k GNU/Linux
PS it seems LVM is working with this kernel as well!
Christian
[ 7.000000] scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq 15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { }, options { REAL_DMA SUPPORT_TAGS }
Checksum status : valid
Boot preference : unspecified
SCSI arbitration : on
SCSI host ID : 7
OS language : English (US)
Keyboard language: English (UK)
Date format : YY?DD?MM, 24h clock
Boot delay : 10s
Video mode : ?? colors, 80 columns, VGA PAL monitor
overscan, compat. mode off, line doubling
00000000 00 00 ff ff ff ff 00 03 13 ff 0a ff ff ff 01 7f |................|
00000010 87 ff 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000030 e1 1e |..|
00000032
3b5801864975cf23bcacb52f648e74cc /dev/nvram
3b5801864975cf23bcacb52f648e74cc /tmp/nvram
Thanks for helping with this, Christian. I'll add your name in "Tested-by"
tags on the relevant patches. These are rudimentary tests but combined
with my own testing on m68k, ppc32 and x86, coverage is quite good. Some
testing on ppc64 is still lacking though.
--
From: Michael Schmitz <schmitzmic@gmail.com> Date: 2015-07-25 00:51:34
Hi Christian,
good to know this worked - for the record (Finn), this is the kernel
with Finn's patch applied.
I'll build the missing LVM module so Christian can run the tests on the
unpatched kernel as well.
Thanks for testing!
Michael
Am 25.07.15 um 07:07 schrieb Christian T. Steigies:
Moin,
On Fri, Jul 24, 2015 at 02:56:26PM +1200, Michael Schmitz wrote:
quoted
here's what Finn asked me to run as tests:
# dmesg | grep this_id > nvram.out
# cat /proc/driver/nvram >> nvram.out
# hexdump -C /dev/nvram >> nvram.out
# cp /dev/nvram /tmp/nvram
# cp /tmp/nvram /dev/nvram
# md5sum /dev/nvram /tmp/nvram >> nvram.out
It seems I could successfully reset the nvram under TOS with bootconf.
Here is nvram.out, using the new kernel you sent me:
root@garkin:~>uname -a
Linux garkin 4.2.0-rc2-atari-269994-gc1e9d12 #327 Fri Jul 24 19:34:44 NZST 2015 m68k GNU/Linux
PS it seems LVM is working with this kernel as well!
Christian
[ 7.000000] scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq 15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { }, options { REAL_DMA SUPPORT_TAGS }
Checksum status : valid
Boot preference : unspecified
SCSI arbitration : on
SCSI host ID : 7
OS language : English (US)
Keyboard language: English (UK)
Date format : YYÿDDÿMM, 24h clock
Boot delay : 10s
Video mode : ?? colors, 80 columns, VGA PAL monitor
overscan, compat. mode off, line doubling
00000000 00 00 ff ff ff ff 00 03 13 ff 0a ff ff ff 01 7f |................|
00000010 87 ff 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000030 e1 1e |..|
00000032
3b5801864975cf23bcacb52f648e74cc /dev/nvram
3b5801864975cf23bcacb52f648e74cc /tmp/nvram
Thanks for helping with this, Christian. I'll add your name in "Tested-by"
tags on the relevant patches. These are rudimentary tests but combined
with my own testing on m68k, ppc32 and x86, coverage is quite good. Some
testing on ppc64 is still lacking though.
*Blush*
I'll try and get to it next week promise!
cheers
From: Finn Thain <hidden> Date: 2015-07-25 07:28:20
On Sat, 25 Jul 2015, Michael Schmitz wrote:
Hi Christian,
good to know this worked - for the record (Finn), this is the kernel
with Finn's patch applied.
That was v5 of this patch series. I will send that out to the lists now.
It has some minor changes that relate to powerpc.
Thanks for your help with this, Michael.
Finn
I'll build the missing LVM module so Christian can run the tests on the
unpatched kernel as well.
Thanks for testing!
Michael
From: Finn Thain <hidden> Date: 2015-07-25 07:38:25
On Sat, 25 Jul 2015, Michael Ellerman wrote:
On Sat, 2015-07-25 at 10:35 +1000, Finn Thain wrote:
quoted
Thanks for helping with this, Christian. I'll add your name in
"Tested-by" tags on the relevant patches. These are rudimentary tests
but combined with my own testing on m68k, ppc32 and x86, coverage is
quite good. Some testing on ppc64 is still lacking though.
*Blush*
I'll try and get to it next week promise!
That would be a great help. Unfortunately I don't have any ppc64 hardware.
I wish ;-)
--
From: Michael Schmitz <schmitzmic@gmail.com> Date: 2015-07-26 01:03:01
Hi Finn,
For the sake of completeness: further testing on ARAnyM shows no
difference between original and patched kernel in the NVRAM proc and
diff outputs:
scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq
15, can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { },
options { REAL_DMA SUPPORT_TAGS }
Checksum status : valid
Boot preference : unspecified
SCSI arbitration : on
SCSI host ID : 7
OS language : English (US)
Keyboard language: English (US)
Date format : DD.MM.YY, 24h clock
Boot delay : 32s
Video mode : 4 colors, 40 columns, TV NTSC monitor
no overscan, compat. mode off
00000000 00 00 00 00 00 00 00 00 11 2e 20 01 ff 00 00 3b |..........
....;|
00000010 87 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
|................|
00000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
|................|
00000030 de 21 |.!|
00000032
d640bf7d535b54e39582fabdc016d7ca /dev/nvram
d640bf7d535b54e39582fabdc016d7ca /tmp/nvram
Happy to help ...
Cheers,
Michael
On Sat, 25 Jul 2015, Michael Schmitz wrote:
quoted
Hi Christian,
good to know this worked - for the record (Finn), this is the kernel
with Finn's patch applied.
That was v5 of this patch series. I will send that out to the lists now.
It has some minor changes that relate to powerpc.
Thanks for your help with this, Michael.
Finn
quoted
I'll build the missing LVM module so Christian can run the tests on the
unpatched kernel as well.
Thanks for testing!
Michael
From: Finn Thain <hidden> Date: 2015-07-26 01:20:07
On Sun, 26 Jul 2015, Michael Schmitz wrote:
Hi Finn,
For the sake of completeness: further testing on ARAnyM shows no difference
between original and patched kernel in the NVRAM proc and diff outputs:
scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq 15,
can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { }, options {
REAL_DMA SUPPORT_TAGS }
Checksum status : valid
Boot preference : unspecified
SCSI arbitration : on
SCSI host ID : 7
OS language : English (US)
Keyboard language: English (US)
Date format : DD.MM.YY, 24h clock
Boot delay : 32s
Video mode : 4 colors, 40 columns, TV NTSC monitor
no overscan, compat. mode off
00000000 00 00 00 00 00 00 00 00 11 2e 20 01 ff 00 00 3b |.......... ....;|
00000010 87 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000030 de 21 |.!|
00000032
I'd expect to see a change in the Video mode above, because of patch 2.
With 0x3B in byte 15, I'd expect the patched kernel to produce something
similar to Christian's /proc/driver/nvram results.
--
From: Finn Thain <hidden> Date: 2015-07-26 01:37:49
On Sat, 25 Jul 2015, Michael Ellerman wrote:
On Sat, 2015-07-25 at 10:35 +1000, Finn Thain wrote:
quoted
... These are rudimentary tests but combined with my own testing on
m68k, ppc32 and x86, coverage is quite good. Some testing on ppc64 is
still lacking though.
Here's some code I wrote some time ago to help me test these patches,
maybe it will help others (though independent results may be more
valuable...)
-- -- 8< -- --
#!/bin/sh
set -e -u -x
temp=`mktemp /root/nvram.XXXX`
test -f "$temp"
cat /proc/version
if ! test -e /dev/nvram ; then
mknod /dev/nvram c 10 144
fi
ls -l /dev/nvram
if test -c /dev/nvram ; then
cp /dev/nvram "$temp"
hexdump -C "$temp"
cp /dev/zero /dev/nvram || true
hexdump -C /dev/nvram
cp "$temp" /dev/nvram
hexdump -C /dev/nvram
fi
-- -- 8< -- --
#include <stdio.h>
#include <sys/ioctl.h>
#include <errno.h>
#if defined(__powerpc__)
#include <asm/nvram.h>
#elif defined(__i386__) || defined(__mc68020__)
#include <linux/nvram.h>
#endif
int main(void) {
int result, i;
FILE *f;
f = fopen("/dev/nvram", "r+");
if (!f) {
perror("fopen");
return 1;
}
#if defined(__powerpc__)
result = ioctl(fileno(f), IOC_NVRAM_SYNC);
fprintf(stderr, "ioctl IOC_NVRAM_SYNC: result %d, errno %m\n", result);
for (i = 0; i <= 8; ++i) {
int arg = i;
result = ioctl(fileno(f), IOC_NVRAM_GET_OFFSET, (long)&arg);
fprintf(stderr, "ioctl IOC_NVRAM_GET_OFFSET: i %d, result %d, offset %d, errno %m\n", i, result, arg);
}
#elif defined(__i386__) || defined(__mc68020__)
result = ioctl(fileno(f), NVRAM_INIT);
fprintf(stderr, "ioctl NVRAM_INIT: result %d, errno %m\n", result);
result = ioctl(fileno(f), NVRAM_SETCKS);
fprintf(stderr, "ioctl NVRAM_SETCKS: result %d, errno %m\n", result);
#endif
return 0;
}
--
@@ -7,7 +7,7 @@ Keyboard language: English (US) Date format : DD.MM.YY, 24h clock Boot delay : 32s-Video mode : 4 colors, 40 columns, TV NTSC monitor+Video mode : 256 colors, 80 columns, VGA PAL monitor no overscan, compat. mode off 00000000 00 00 00 00 00 00 00 00 11 2e 20 01 ff 00 00 3b |.......... ....;| 00000010 84 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
That video mode is indeed the one set in the NVRAM (by ARAnyM config,
if running emulated)
Cheers,
Michael
On Sun, Jul 26, 2015 at 1:19 PM, Finn Thain [off-list ref] wrote:
On Sun, 26 Jul 2015, Michael Schmitz wrote:
quoted
Hi Finn,
For the sake of completeness: further testing on ARAnyM shows no difference
between original and patched kernel in the NVRAM proc and diff outputs:
scsi host0: Atari native SCSI, io_port 0x0, n_io_port 0, base 0x0, irq 15,
can_queue 8, cmd_per_lun 1, sg_tablesize 0, this_id 7, flags { }, options {
REAL_DMA SUPPORT_TAGS }
Checksum status : valid
Boot preference : unspecified
SCSI arbitration : on
SCSI host ID : 7
OS language : English (US)
Keyboard language: English (US)
Date format : DD.MM.YY, 24h clock
Boot delay : 32s
Video mode : 4 colors, 40 columns, TV NTSC monitor
no overscan, compat. mode off
00000000 00 00 00 00 00 00 00 00 11 2e 20 01 ff 00 00 3b |.......... ....;|
00000010 87 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
00000030 de 21 |.!|
00000032
I'd expect to see a change in the Video mode above, because of patch 2.
With 0x3B in byte 15, I'd expect the patched kernel to produce something
similar to Christian's /proc/driver/nvram results.
--