Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
(PReP with no RTAS but with OF).
I use netboot to test new kernels. Normally, the screen is changed to
VGA text mode and the bootloader speaks some lines of info and asks for
the kernel command line. Now, the image is loaded via tftp (as shown by
tcpdump, the last datagram is smaller) and nothing more happens. The
cursor stays where it is - at the beginning of the Booting ... line in
graphics mode OF environment and that's all.
--
Meelis Roos ([email protected])
On Fri, Mar 26, 2004 at 08:33:18PM +0200, Meelis Roos wrote:
> Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
> (PReP with no RTAS but with OF).
>
> I use netboot to test new kernels. Normally, the screen is changed to
> VGA text mode and the bootloader speaks some lines of info and asks for
> the kernel command line. Now, the image is loaded via tftp (as shown by
> tcpdump, the last datagram is smaller) and nothing more happens. The
> cursor stays where it is - at the beginning of the Booting ... line in
> graphics mode OF environment and that's all.
Hmm. Can you hook up a serial line, enable serial console as well and
see if anything pops out there?
--
Tom Rini
http://gate.crashing.org/~trini/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Meelis Roos wrote:
| Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
| (PReP with no RTAS but with OF).
|
| I use netboot to test new kernels. Normally, the screen is changed to
| VGA text mode and the bootloader speaks some lines of info and asks for
| the kernel command line. Now, the image is loaded via tftp (as shown by
| tcpdump, the last datagram is smaller) and nothing more happens. The
| cursor stays where it is - at the beginning of the Booting ... line in
| graphics mode OF environment and that's all.
|
are you sure this is not the issue "Blank screen after decompressing
kernel" described here:
http://www.codemonkey.org.uk/post-halloween-2.5.txt
| Make sure your .config has
| CONFIG_INPUT=y
| CONFIG_VT=y
| CONFIG_VGA_CONSOLE=y
| CONFIG_VT_CONSOLE=y
i really like to hear what your PowerStack is doing after booting. i
still have severe issues with 2.5/2.6 on my PReP...
Christian.
- --
BOFH excuse #227:
Fatal error right in front of screen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFAZKac+A7rjkF8z0wRAp+xAJ9oDHmbPcTaL9SuEGdQNVWZI/1S/QCfbiO2
UdcbbEZnXyFClAmm4t/WGgY=
=WIXh
-----END PGP SIGNATURE-----
Tom Rini <[email protected]> wrote:
> On Fri, Mar 26, 2004 at 08:33:18PM +0200, Meelis Roos wrote:
>> Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
>> (PReP with no RTAS but with OF).
>>
>> I use netboot to test new kernels. Normally, the screen is changed to
>> VGA text mode and the bootloader speaks some lines of info and asks for
>> the kernel command line. Now, the image is loaded via tftp (as shown by
>> tcpdump, the last datagram is smaller) and nothing more happens. The
>> cursor stays where it is - at the beginning of the Booting ... line in
>> graphics mode OF environment and that's all.
> Hmm. Can you hook up a serial line, enable serial console as well and
> see if anything pops out there?
I am seeing the same problem here too (Powerstack II with OF).
Using the serial console does not change to behavior in any way, it is
just stuck after the transmission of the kernel and no further
information is printed.
I even let it stay this way for a long time (read: over 12 hours) and
nothing changed.
The only difference is: Is see this problem since the 2.6.4 (or maybe
even 2.6.3) kernel. I'm a bit short on time right now, so I am
currently not able to specify a correct -BK-release.
At least I know, 2.6.1 and 2.6.2-rc2 were able to boot, until the
problem with the Tulip/3Com cards hits me
(http://bugzilla.kernel.org/show_bug.cgi?id=1494)
Gr??e,
S?
--
Fachbegriffe der Informatik - Einfach erkl?rt
91: Emacs
Eight Megabytes And Continously Swapping
Christian Kujau <[email protected]> wrote:
> Meelis Roos wrote:
>| Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
>| (PReP with no RTAS but with OF).
>|
>| I use netboot to test new kernels. Normally, the screen is changed to
>| VGA text mode and the bootloader speaks some lines of info and asks for
>| the kernel command line. Now, the image is loaded via tftp (as shown by
>| tcpdump, the last datagram is smaller) and nothing more happens. The
>| cursor stays where it is - at the beginning of the Booting ... line in
>| graphics mode OF environment and that's all.
> are you sure this is not the issue "Blank screen after decompressing
> kernel" described here:
> http://www.codemonkey.org.uk/post-halloween-2.5.txt
I don't think so, since 2.6.1 and 2.6.2-rc2 are booting just fine (if I
don't plan to use the onboard nic).
Gr??e,
S?
--
Letzte Worte eines Kannibalen: Was ist eigentlich ein Androide?
On Sat, Mar 27, 2004 at 02:55:09PM +0100, Sven Hartge wrote:
> Tom Rini <[email protected]> wrote:
> > On Fri, Mar 26, 2004 at 08:33:18PM +0200, Meelis Roos wrote:
>
> >> Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
> >> (PReP with no RTAS but with OF).
> >>
> >> I use netboot to test new kernels. Normally, the screen is changed to
> >> VGA text mode and the bootloader speaks some lines of info and asks for
> >> the kernel command line. Now, the image is loaded via tftp (as shown by
> >> tcpdump, the last datagram is smaller) and nothing more happens. The
> >> cursor stays where it is - at the beginning of the Booting ... line in
> >> graphics mode OF environment and that's all.
>
> > Hmm. Can you hook up a serial line, enable serial console as well and
> > see if anything pops out there?
>
> I am seeing the same problem here too (Powerstack II with OF).
>
> Using the serial console does not change to behavior in any way, it is
> just stuck after the transmission of the kernel and no further
> information is printed.
OK. I _think_ I know where the problem is, and I'll try and look into
it today.
--
Tom Rini
http://gate.crashing.org/~trini/
On Sat, Mar 27, 2004 at 02:55:09PM +0100, Sven Hartge wrote:
> Tom Rini <[email protected]> wrote:
> > On Fri, Mar 26, 2004 at 08:33:18PM +0200, Meelis Roos wrote:
>
> >> Recent 2.6.5-pre* and -rc1 and -BK don't boot on my Motorola Powerstack
> >> (PReP with no RTAS but with OF).
> >>
> >> I use netboot to test new kernels. Normally, the screen is changed to
> >> VGA text mode and the bootloader speaks some lines of info and asks for
> >> the kernel command line. Now, the image is loaded via tftp (as shown by
> >> tcpdump, the last datagram is smaller) and nothing more happens. The
> >> cursor stays where it is - at the beginning of the Booting ... line in
> >> graphics mode OF environment and that's all.
>
> > Hmm. Can you hook up a serial line, enable serial console as well and
> > see if anything pops out there?
>
> I am seeing the same problem here too (Powerstack II with OF).
Ok. Can both of you try the following patch on top of the version which
fails?
arch/ppc/boot/simple/head.S | 2 +-
arch/ppc/boot/simple/misc-prep.c | 4 +++-
2 files changed, 4 insertions(+), 2 deletions(-)
--- 1.8/arch/ppc/boot/simple/head.S Mon Mar 1 16:34:28 2004
+++ edited/arch/ppc/boot/simple/head.S Mon Mar 29 08:14:48 2004
@@ -99,8 +99,8 @@
mr r25,r5
#else
bl disable_6xx_mmu
-#endif
bl disable_6xx_l1cache
+#endif
CLEAR_CACHES
#endif
--- 1.1/arch/ppc/boot/simple/misc-prep.c Mon Mar 1 16:34:28 2004
+++ edited/arch/ppc/boot/simple/misc-prep.c Mon Mar 29 08:13:08 2004
@@ -35,6 +35,7 @@
extern struct bi_record *decompress_kernel(unsigned long load_addr,
int num_words, unsigned long cksum);
extern void disable_6xx_mmu(void);
+extern void disable_6xx_l1cache(void);
extern unsigned long mpc10x_get_mem_size(void);
static void
@@ -152,8 +153,9 @@
}
/* Now go and clear out the BATs and ensure that our MSR is
- * correct .*/
+ * correct. Also disable the L1s finally. */
disable_6xx_mmu();
+ disable_6xx_l1cache();
/* Make r3 be a pointer to the residual data. */
return (unsigned long)hold_residual;
--
Tom Rini
http://gate.crashing.org/~trini/
> Ok. Can both of you try the following patch on top of the version which
> fails?
Tried it on top of fresh 2.6.5-rc3, no changes, it still hangs.
--
Meelis Roos ([email protected])
Meelis Roos <[email protected]> wrote:
>> Ok. Can both of you try the following patch on top of the version
>> which fails?
> Tried it on top of fresh 2.6.5-rc3, no changes, it still hangs.
Same here, still totally dead after tftp.
S?
--
BOFH excuse #9:
doppler effect
Um 08:15 Uhr am 29.03.04 schrieb Tom Rini:
> Ok. Can both of you try the following patch on top of the version which
> fails?
2.6.4 and earlier are unpatchable, because they differ to much from 2.6.5.
Someone with a better connection than a tiny ISDN line should check the
ppc32-changes between 2.6.2-rc2 and 2.6.4, because 2.6.2-rc2 was the last
kernel I tried before 2.6.4 and 2.6.2-rc2 did at least boot.
2.6.5-rc2 and -rc3 don't behave any better after patching than before.
S?
On Tue, Mar 30, 2004 at 08:49:15PM +0200, Sven Hartge wrote:
> Um 08:15 Uhr am 29.03.04 schrieb Tom Rini:
>
> > Ok. Can both of you try the following patch on top of the version which
> > fails?
>
> 2.6.4 and earlier are unpatchable, because they differ to much from 2.6.5.
> Someone with a better connection than a tiny ISDN line should check the
> ppc32-changes between 2.6.2-rc2 and 2.6.4, because 2.6.2-rc2 was the last
> kernel I tried before 2.6.4 and 2.6.2-rc2 did at least boot.
I know what the failure is due to, in general terms. It's merging
arch/ppc/boot/prep/ into arch/ppc/boot/simple/ . What I'm not so sure
about are the specific breakage, but I know it has to do with the
OpenFirmware code as it does work (or rather, it did when I submitted
it, and I'm going to verify that 2.6.5-rc3 stock does today) on my
Motorola PowerStack Series E (PPC1Bug, not OpenFirmware).
--
Tom Rini
http://gate.crashing.org/~trini/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
[ cc'ing linuxppc-dev ]
Sven Hartge wrote:
| Meelis Roos <[email protected]> wrote:
|
|
|>>Ok. Can both of you try the following patch on top of the version
|>>which fails?
|
|
|>Tried it on top of fresh 2.6.5-rc3, no changes, it still hangs.
|
|
| Same here, still totally dead after tftp.
not so dead here. 2.6.4 is ok, 2.6.5-rc1|2|3 are loaded within the OF
menu, but no bootprompt appears. but: i can hear the scsi disk
initalizing, short after this, the atkbd is recognized and the LEDs on
my keyboard are flashing. then again my nfs-root is supposed to be
mounted, but my PReP still locks up completely upon network-init. (last
working is still 2.5.30).
another issue here: i was finally able to cross-compile 2.5.x / 2.6.x
kernels (on x86). i tried to compile kernels from 2.5.21 on with
"allnoconfig" (was introduced in 2.5.21). only 2.5.30 can be built, all
other attempts to build "zImage" fail...(still compiling 2.5.6x)...
(full logs of builds available...)
Christian.
- --
BOFH excuse #204:
Just pick up the phone and give modem connect sounds. "Well you said we
should get more lines so we don't have voice lines."
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFAcEdC+A7rjkF8z0wRAq5TAJsHIh5V7wb/IP1xW7uHde4nC3EquACgk4D+
TLrtSsdwpdtZBXCRmD9fiE4=
=aGiL
-----END PGP SIGNATURE-----
On Sun, Apr 04, 2004 at 07:34:59PM +0200, Christian Kujau wrote:
> [ cc'ing linuxppc-dev ]
>
> Sven Hartge wrote:
> | Meelis Roos <[email protected]> wrote:
> |
> |
> |>>Ok. Can both of you try the following patch on top of the version
> |>>which fails?
> |
> |
> |>Tried it on top of fresh 2.6.5-rc3, no changes, it still hangs.
> |
> |
> | Same here, still totally dead after tftp.
>
> not so dead here. 2.6.4 is ok, 2.6.5-rc1|2|3 are loaded within the OF
> menu, but no bootprompt appears. but: i can hear the scsi disk
> initalizing, short after this, the atkbd is recognized and the LEDs on
> my keyboard are flashing. then again my nfs-root is supposed to be
> mounted, but my PReP still locks up completely upon network-init. (last
> working is still 2.5.30).
OK, hmm. I've got some better ideas then. It sounds like the code to
have puts show up on VGA isn't selected/compiled in. Or, there's still
some other problem wrt the OF transition code. Just having a serial
console selected still doesn't give output however, right?
> another issue here: i was finally able to cross-compile 2.5.x / 2.6.x
> kernels (on x86). i tried to compile kernels from 2.5.21 on with
> "allnoconfig" (was introduced in 2.5.21). only 2.5.30 can be built, all
> other attempts to build "zImage" fail...(still compiling 2.5.6x)...
> (full logs of builds available...)
The simple answer is, don't use allnoconfig :). Do a 'make
common_defconfig' and then from there turn off stuff you don't need.
--
Tom Rini
http://gate.crashing.org/~trini/
On Sun, Apr 04, 2004 at 07:34:59PM +0200, Christian Kujau wrote:
> [ cc'ing linuxppc-dev ]
>
> Sven Hartge wrote:
> | Meelis Roos <[email protected]> wrote:
> |
> |
> |>>Ok. Can both of you try the following patch on top of the version
> |>>which fails?
> |
> |
> |>Tried it on top of fresh 2.6.5-rc3, no changes, it still hangs.
> |
> |
> | Same here, still totally dead after tftp.
>
> not so dead here. 2.6.4 is ok, 2.6.5-rc1|2|3 are loaded within the OF
> menu, but no bootprompt appears. but: i can hear the scsi disk
> initalizing, short after this, the atkbd is recognized and the LEDs on
> my keyboard are flashing. then again my nfs-root is supposed to be
> mounted, but my PReP still locks up completely upon network-init. (last
> working is still 2.5.30).
OK, hmm. I've got some better ideas then. It sounds like the code to
have puts show up on VGA isn't selected/compiled in. Or, there's still
some other problem wrt the OF transition code. Just having a serial
console selected still doesn't give output however, right?
> another issue here: i was finally able to cross-compile 2.5.x / 2.6.x
> kernels (on x86). i tried to compile kernels from 2.5.21 on with
> "allnoconfig" (was introduced in 2.5.21). only 2.5.30 can be built, all
> other attempts to build "zImage" fail...(still compiling 2.5.6x)...
> (full logs of builds available...)
The simple answer is, don't use allnoconfig :). Do a 'make
common_defconfig' and then from there turn off stuff you don't need.
--
Tom Rini
http://gate.crashing.org/~trini/
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
[ if someone is bored from me cc'ing too many - plz cry! ]
Tom Rini wrote:
| OK, hmm. I've got some better ideas then. It sounds like the code to
| have puts show up on VGA isn't selected/compiled in. Or, there's still
| some other problem wrt the OF transition code. Just having a serial
| console selected still doesn't give output however, right?
I'll give it a try with 2.6 this week.
|
|>another issue here: i was finally able to cross-compile 2.5.x / 2.6.x
|>kernels (on x86). i tried to compile kernels from 2.5.21 on with
|>"allnoconfig" (was introduced in 2.5.21). only 2.5.30 can be built, all
|>other attempts to build "zImage" fail...(still compiling 2.5.6x)...
|>(full logs of builds available...)
|
| The simple answer is, don't use allnoconfig :). Do a 'make
| common_defconfig' and then from there turn off stuff you don't need.
um, yes. but the target "common_defconfig" was disabled somewhere in
2.5, so my shini script broke. i wanted to do common_defconfig first,
then always keep my .config and do "oldconfig" after patching, but
somehow my script broke, so i went with "allnoconfig"...but ok, i'll try
again.
if anyone is interested: http://nerdbynature.de/bits/sheep/build/
all the logfiles produced when patchin/compiling 2.5.21 up to 2.5.75.
right here the latest 2.4-benh and 2.6-benh kernels (via rsync), also
compiled via crosscompile: http://nerdbynature.de/bits/sheep/latest-kernel/
thanks,
Christian.
- --
BOFH excuse #212:
Of course it doesn't work. We've performed a software upgrade.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFAcc1Q+A7rjkF8z0wRAtxBAKCrbpGe3/HlW8UMLAdFKHCnWiyB/QCfV4FC
hVoNM7pW01ntTdFBETVxT44=
=jCF2
-----END PGP SIGNATURE-----
[ if someone is bored from me cc'ing too many - plz cry! ]
Tom Rini wrote:
| OK, hmm. I've got some better ideas then. It sounds like the code to
| have puts show up on VGA isn't selected/compiled in. Or, there's still
| some other problem wrt the OF transition code. Just having a serial
| console selected still doesn't give output however, right?
I'll give it a try with 2.6 this week.
|
|>another issue here: i was finally able to cross-compile 2.5.x / 2.6.x
|>kernels (on x86). i tried to compile kernels from 2.5.21 on with
|>"allnoconfig" (was introduced in 2.5.21). only 2.5.30 can be built, all
|>other attempts to build "zImage" fail...(still compiling 2.5.6x)...
|>(full logs of builds available...)
|
| The simple answer is, don't use allnoconfig :). Do a 'make
| common_defconfig' and then from there turn off stuff you don't need.
um, yes. but the target "common_defconfig" was disabled somewhere in
2.5, so my shini script broke. i wanted to do common_defconfig first,
then always keep my .config and do "oldconfig" after patching, but
somehow my script broke, so i went with "allnoconfig"...but ok, i'll try
again.
if anyone is interested: http://nerdbynature.de/bits/sheep/build/
all the logfiles produced when patchin/compiling 2.5.21 up to 2.5.75.
right here the latest 2.4-benh and 2.6-benh kernels (via rsync), also
compiled via crosscompile: http://nerdbynature.de/bits/sheep/latest-kernel/
thanks,
Christian.
- --
BOFH excuse #212:
Of course it doesn't work. We've performed a software upgrade.
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
On Monday 05 April 2004 16:19, Christian Kujau wrote:
> um, yes. but the target "common_defconfig" was disabled somewhere in
> 2.5, so my shini script broke. i wanted to do common_defconfig first,
> then always keep my .config and do "oldconfig" after patching, but
> somehow my script broke, so i went with "allnoconfig"...but ok, i'll try
> again.
FWIW, I use ibmchrp_config and then enable a couple PReP-specific things. It's
pretty bare...
--
Hollis Blanchard
IBM Linux Technology Center
On Monday 05 April 2004 16:19, Christian Kujau wrote:
> um, yes. but the target "common_defconfig" was disabled somewhere in
> 2.5, so my shini script broke. i wanted to do common_defconfig first,
> then always keep my .config and do "oldconfig" after patching, but
> somehow my script broke, so i went with "allnoconfig"...but ok, i'll try
> again.
FWIW, I use ibmchrp_config and then enable a couple PReP-specific things. It's
pretty bare...
--
Hollis Blanchard
IBM Linux Technology Center
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/