linuxsh-dev

[SUPERH / PATA / SCSI] Unable to do start userland after kernel boot

[Kristoffer E. <kri...@gm...>]

Greetings,

I know I'm being annoying about this, but kinda a show stopper for hp6xx currently.
I compaired a bootlog (good kernel vs bad kernel), and they are identical apart from this stuff

The bottom line seems to be that it fails to attach scsi sg0. It explains why it doesn't work, but not why it stopped working. And this has nothing to do with the current kernel config, I've been over that the last 4 days. It all started when I synced my jlime-current.git repository with linux-2.6.git. As you can see I had 2.6.23-rc6 and then synced up to 2.6.23-rc8/rc9 and thats when the troubles started.

Any scsi/libata gurus giving feedback on this would be appreciated.

Best wishes
Kristoffer Ericson

[BAD KERNEL]
(Linux version 2.6.23-gbc53c3a1-dirty (kristoffer@Wolverine) (gcc version 3.4.5) #2 Tue Oct 23 19:34:27 PDT 2007)
scsi0 : pata_platform
ata1: PATA max PIO0 mmio cmd 0x150001f0 ctl 0x150001fe irq 77  <-- "mmio cmd" + ""
ata1.00: CFA: Hitachi XX.V.3.5.0.0, Rev 0.00, max PIO4
ata1.00: 2002896 sectors, multi 0: LBA
ata1.00: configured for PIO
scsi 0:0:0:0: Direct-Access     ATA      Hitachi XX.V.3.5 Rev  PQ: 0 ANSI: 5
sd 0:0:0:0: [sda] 2002896 512-byte hardware sectors (1025 MB)
sd 0:0:0:0: [sda] Write Protect is off
sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
sd 0:0:0:0: [sda] 2002896 512-byte hardware sectors (1025 MB)
sd 0:0:0:0: [sda] Write Protect is off
sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
sda: sda1 sda2 sda3
sd 0:0:0:0: [sda] Attached SCSI removable disk
.........
.........
........
EXT2-fs warning: mounting unchecked fs, running e2fsck is recommended
VFS: Mounted root (ext2 filesystem).
Freeing unused kernel memory: 128k freed
Failed to execute /bin/sh.  Attempting defaults...
Kernel panic - not syncing: No init found.  Try passing init= option to kernel.


[GOOD KERNEL]
(Linux version 2.6.23-rc6-gab762e2c-dirty (rafa@nodo3) (gcc version 3.4.4) #51 Sun Oct 21 15:23:19 ART 2007)
scsi0 : pata_platform
ata1: PATA max PIO0 cmd 0xb50001f0 ctl 0xb50001fe bmdma 0x00000000 irq 77 <--- "cmd" + "bmdma"
ata1.00: CFA: Hitachi XX.V.3.5.0.0, Rev 0.00, max PIO4
ata1.00: 2002896 sectors, multi 0: LBA
ata1.00: configured for PIO
ata1: EH pending after completion, repeating EH (cnt=4)
scsi 0:0:0:0: Direct-Access     ATA      Hitachi XX.V.3.5 Rev  PQ: 0 ANSI: 5
sd 0:0:0:0: [sda] 2002896 512-byte hardware sectors (1025 MB)
sd 0:0:0:0: [sda] Write Protect is off
sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
sd 0:0:0:0: [sda] 2002896 512-byte hardware sectors (1025 MB)
sd 0:0:0:0: [sda] Write Protect is off
sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
sda: sda1 sda2 sda3
sd 0:0:0:0: [sda] Attached SCSI removable disk
sd 0:0:0:0: Attached scsi generic sg0 type 0   <---- Only here, missing from bad kernel.
............
...........
.............
EXT2-fs warning: mounting unchecked fs, running e2fsck is recommended
VFS: Mounted root (ext2 filesystem).
Freeing unused kernel memory: 140k freed
/bin/sh: can't access tty; job control turned off       (init=/bin/sh was used as bootparam)
/ $ 

Re: [SUPERH / PATA / SCSI] Unable to do start userland after kernel boot

[Paul M. <le...@li...>]

On Thu, Oct 25, 2007 at 09:22:40PM -0700, Kristoffer Ericson wrote:
> The bottom line seems to be that it fails to attach scsi sg0. It
> explains why it doesn't work, but not why it stopped working. And this
> has nothing to do with the current kernel config, I've been over that
> the last 4 days. It all started when I synced my jlime-current.git
> repository with linux-2.6.git. As you can see I had 2.6.23-rc6 and then
> synced up to 2.6.23-rc8/rc9 and thats when the troubles started.
> 
At least that suggests it's not fallout from the INTC changes in -rc1, so
that helps to narrow it down a bit. Since you have a known good and bad,
it would be nice if you could bisect this to figure out what exactly
caused the regression. There weren't any SH-specific changes between rc6
and rc8/rc9 at least.

Re: [SUPERH / PATA / SCSI] Unable to do start userland after kernel boot

[Kristoffer E. <kri...@gm...>]

Greetings,

I made another BAD kernel just to make sure that generic was inside it,
turns out the new bad kernel gives out : sd 0:0:0:0: Attached scsi generic
sg0 type 0 and still fails.
My mistake for not compiling the latest bad kernel myself.
So back to square one then I guess.

I did a small bisect last night going through arch/sh only and it didn't
produce any workable kernel, so atleast that implies the error lies
elsewhere.
Will see if I can bisect all changes and see where I end up.

Though:
ata1: PATA max PIO0 mmio cmd 0x150001f0 ctl 0x150001fe irq 77 this line
still contains "+mmio" when compaired to the good kernel. Not sure if that
actually reflects anything since we were using mmio long before, but perhaps
didn't inform about it at bootup.

Thanks for commenting paul

2007/10/25, Paul Mundt <le...@li...>:
>
> On Thu, Oct 25, 2007 at 09:22:40PM -0700, Kristoffer Ericson wrote:
> > The bottom line seems to be that it fails to attach scsi sg0. It
> > explains why it doesn't work, but not why it stopped working. And this
> > has nothing to do with the current kernel config, I've been over that
> > the last 4 days. It all started when I synced my jlime-current.git
> > repository with linux-2.6.git. As you can see I had 2.6.23-rc6 and then
> > synced up to 2.6.23-rc8/rc9 and thats when the troubles started.
> >
> At least that suggests it's not fallout from the INTC changes in -rc1, so
> that helps to narrow it down a bit. Since you have a known good and bad,
> it would be nice if you could bisect this to figure out what exactly
> caused the regression. There weren't any SH-specific changes between rc6
> and rc8/rc9 at least.
>

Re: [SUPERH / PATA / SCSI] Unable to do start userland after kernel boot (BISECTED)

[Kristoffer E. <kri...@gm...>]

On Fri, 26 Oct 2007 04:46:18 +0900
Paul Mundt <le...@li...> wrote:

> On Thu, Oct 25, 2007 at 09:22:40PM -0700, Kristoffer Ericson wrote:
> > The bottom line seems to be that it fails to attach scsi sg0. It
> > explains why it doesn't work, but not why it stopped working. And this
> > has nothing to do with the current kernel config, I've been over that
> > the last 4 days. It all started when I synced my jlime-current.git
> > repository with linux-2.6.git. As you can see I had 2.6.23-rc6 and then
> > synced up to 2.6.23-rc8/rc9 and thats when the troubles started.
> > 
> At least that suggests it's not fallout from the INTC changes in -rc1, so
> that helps to narrow it down a bit. Since you have a known good and bad,
> it would be nice if you could bisect this to figure out what exactly
> caused the regression. There weren't any SH-specific changes between rc6
> and rc8/rc9 at least.

I've spent a couple of hours bisecting it, and this is where it stops working. I haven't reversed the patch yet, simply because Im dead tired :D Will do that tommorow, but Im quite confident that this is the correct bug.

023ef184fff6ac2e7cba345708f35536a2a419cb is first bad commit
commit 023ef184fff6ac2e7cba345708f35536a2a419cb
Author: Stuart Menefy <stu...@st...>
Date:   Fri Sep 28 12:36:35 2007 +0900

    sh: __copy_user() optimizations for small copies.
    
    This implements a fast-path for small (less than 12 bytes) copies,
    with the existing path treated as the slow-path and left as the default
    behaviour for all other copy sizes.
    
    Signed-off-by: Stuart Menefy <stu...@st...>
    Signed-off-by: Paul Mundt <le...@li...>

:040000 040000 43f62cf05d1f71a5564b232dfd9e8492af909a90 4ab51dc5b85bc9bc86d58331845e525a67751be8

My bisect log:
*START_ BAD     -   _CURRENT_                                                   30 October              "No Init found"
BAD     -   b5869ce7f68b233ceb81465a7655be0d9a5f3dbb    "Merge git://..sched"   15 October              "No Init found"
BAD     -   f248488b397d52717f6683e2e53200aa687ffc89    "merge infradead.org"   14 October              "No Init found"
BAD     -   3749c66c67fb5c257771815c186bc32290cacf44    "merge git/avi/kvm"     13 October              "No Init found"
BAD     -   dcf397f037f52add9945eced57ca300ab6a4413c    "merge sh-2.6"          13 October              "No Init found"
BAD     -   5d9df8eeacec943c9599f1cfd1069bc8cced3de6    "sh: Fix SH-4 DMAC.."   "8 October"             "No Init found"
BAD     -   e5137682a1ad48bc5306070935c277e262f119ef    "sh: Tidy up gUSA .."   "28 September"          "No Init found"
BAD     -   023ef184fff6ac2e7cba345708f35536a2a419cb    "sh: __copy_user().."   "28 September"          "No Init found"
GOOD    -   24eb17e0813490497f4d5b2fad218bdba402cece    "sh: clkfwk: Support."  "28 September"          "WORKS!"
GOOD    -   cb7af21f7d370edb3a6a6d3e15cb17c8fd61591e    "sh: Use boot_cpu_d."   "27 September"          "WORKS!"
GOOD    -   c167aeef232c45deaf5c6c9be00a1f71b14962d3    "sh: Kill off dupl.."   "27 September"          "WORKS!"
GOOD    -   1db4e9bb5682fd3fd3f37f7fe9c322e7c5bb7578    "sh: don't enable.."    "11 September"          "WORKS!"
GOOD    -   ab9c232286c2b77be78441c2d8396500b045777e    "Merge..libata-dev"     12 October              "WORKS!"
GOOD    -   ce9d3c9a6a9aef61525be07fe6ba27d937236aa2    "Merge-br.. for linux"  11 October              "WORKS!"
GOOD    -   d85f57938ad1d674dff8077a2e6a36a45dbe0e22    "Merge branch 'master'" 26 Sept                 "WORKS!"
GOOD    -   2aee6198652b32e5eaef29a8f8330a9dd15b8efd    "fixes-jgarzik"         25 Sept                 "WORKS!"
GOOD    -   f3d5e3a4155b6f42f6f6f0a2cc95ca0adbabe1af    "[PPP] L2TP: Fix ..     19 Sept                 "WORKS!"
GOOD    -   53a3f3087be361dacfc02e7a85b6d6142a41ce8a    ~2.6.23-rc              14 Sept                 "WORKS!"
GOOD    -   ea3c4b126ad63bd782c7bb5266bb4fd88e203169    ~2.6.23-rc               4 Sept                 "WORKS!"

Re: [SUPERH / PATA / SCSI] Unable to do start userland after kernel boot (BISECTED)

[Kristoffer E. <kri...@gm...>]

On Wed, 31 Oct 2007 02:56:21 -0700
Kristoffer Ericson <Kri...@gm...> wrote:

> On Fri, 26 Oct 2007 04:46:18 +0900
> Paul Mundt <le...@li...> wrote:
> 
> > On Thu, Oct 25, 2007 at 09:22:40PM -0700, Kristoffer Ericson wrote:
> > > The bottom line seems to be that it fails to attach scsi sg0. It
> > > explains why it doesn't work, but not why it stopped working. And this
> > > has nothing to do with the current kernel config, I've been over that
> > > the last 4 days. It all started when I synced my jlime-current.git
> > > repository with linux-2.6.git. As you can see I had 2.6.23-rc6 and then
> > > synced up to 2.6.23-rc8/rc9 and thats when the troubles started.
> > > 
> > At least that suggests it's not fallout from the INTC changes in -rc1, so
> > that helps to narrow it down a bit. Since you have a known good and bad,
> > it would be nice if you could bisect this to figure out what exactly
> > caused the regression. There weren't any SH-specific changes between rc6
> > and rc8/rc9 at least.

Just wanted to confirm that when reverting the patch below my kernels start booting again. Bisect is a pain to go through, but is really effective.

> 
> I've spent a couple of hours bisecting it, and this is where it stops working. I haven't reversed the patch yet, simply because Im dead tired :D Will do that tommorow, but Im quite confident that this is the correct bug.
> 
> 023ef184fff6ac2e7cba345708f35536a2a419cb is first bad commit
> commit 023ef184fff6ac2e7cba345708f35536a2a419cb
> Author: Stuart Menefy <stu...@st...>
> Date:   Fri Sep 28 12:36:35 2007 +0900
> 
>     sh: __copy_user() optimizations for small copies.
>     
>     This implements a fast-path for small (less than 12 bytes) copies,
>     with the existing path treated as the slow-path and left as the default
>     behaviour for all other copy sizes.
>     
>     Signed-off-by: Stuart Menefy <stu...@st...>
>     Signed-off-by: Paul Mundt <le...@li...>
> 
> :040000 040000 43f62cf05d1f71a5564b232dfd9e8492af909a90 4ab51dc5b85bc9bc86d58331845e525a67751be8
> 
> My bisect log:
> *START_ BAD     -   _CURRENT_                                                   30 October              "No Init found"
> BAD     -   b5869ce7f68b233ceb81465a7655be0d9a5f3dbb    "Merge git://..sched"   15 October              "No Init found"
> BAD     -   f248488b397d52717f6683e2e53200aa687ffc89    "merge infradead.org"   14 October              "No Init found"
> BAD     -   3749c66c67fb5c257771815c186bc32290cacf44    "merge git/avi/kvm"     13 October              "No Init found"
> BAD     -   dcf397f037f52add9945eced57ca300ab6a4413c    "merge sh-2.6"          13 October              "No Init found"
> BAD     -   5d9df8eeacec943c9599f1cfd1069bc8cced3de6    "sh: Fix SH-4 DMAC.."   "8 October"             "No Init found"
> BAD     -   e5137682a1ad48bc5306070935c277e262f119ef    "sh: Tidy up gUSA .."   "28 September"          "No Init found"
> BAD     -   023ef184fff6ac2e7cba345708f35536a2a419cb    "sh: __copy_user().."   "28 September"          "No Init found"
> GOOD    -   24eb17e0813490497f4d5b2fad218bdba402cece    "sh: clkfwk: Support."  "28 September"          "WORKS!"
> GOOD    -   cb7af21f7d370edb3a6a6d3e15cb17c8fd61591e    "sh: Use boot_cpu_d."   "27 September"          "WORKS!"
> GOOD    -   c167aeef232c45deaf5c6c9be00a1f71b14962d3    "sh: Kill off dupl.."   "27 September"          "WORKS!"
> GOOD    -   1db4e9bb5682fd3fd3f37f7fe9c322e7c5bb7578    "sh: don't enable.."    "11 September"          "WORKS!"
> GOOD    -   ab9c232286c2b77be78441c2d8396500b045777e    "Merge..libata-dev"     12 October              "WORKS!"
> GOOD    -   ce9d3c9a6a9aef61525be07fe6ba27d937236aa2    "Merge-br.. for linux"  11 October              "WORKS!"
> GOOD    -   d85f57938ad1d674dff8077a2e6a36a45dbe0e22    "Merge branch 'master'" 26 Sept                 "WORKS!"
> GOOD    -   2aee6198652b32e5eaef29a8f8330a9dd15b8efd    "fixes-jgarzik"         25 Sept                 "WORKS!"
> GOOD    -   f3d5e3a4155b6f42f6f6f0a2cc95ca0adbabe1af    "[PPP] L2TP: Fix ..     19 Sept                 "WORKS!"
> GOOD    -   53a3f3087be361dacfc02e7a85b6d6142a41ce8a    ~2.6.23-rc              14 Sept                 "WORKS!"
> GOOD    -   ea3c4b126ad63bd782c7bb5266bb4fd88e203169    ~2.6.23-rc               4 Sept                 "WORKS!"
> 
>