kernel panic with latest kirkwood 3.9 update

[moonman]

I looked some more under https://github.com/torvalds/linux/blob/ ... ug-setup.c
And it seems esata sheevaplug is the one that should not have any problems because both card detect and writeprotect gpios have been initialized, but I think I know why non esata sheevaplug was having problems: static struct mvsdio_platform_data sheevaplug_mvsdio_data hasn't been patched like for other devices. I may need both of you guys (replabrobin and ThisIsAandomString) to test the kernels as I don't have either of these devices and I don't want to fix one and break the other. ThisIsAandomString please see if your card actually isn't writeprotected (or switch it to write protected and see if maybe the functionality is reversed and you can write to a writeprotected card). As a last ditch attempt if everything else fails it should be possible to remove writeprotect checks (this would be a very ugly hack).

I've been looking into the changes: https://github.com/torvalds/linux/commi ... t/mvsdio.c
There were A LOT in 3.9. I found that the patch we are applying now for the sd card to work has actually made it into 3.10.

[ThisIsAandomString]

Hi again.

1 reverted + 1 applied shows the same problem, read only SD card.

Current status for my eSata Sheevaplug:

3.7.3: Fine
3.8.5: Fine
3.8.8: Fine
3.9.2: SD card read-only
3.9.2 patch -1+1: SD card read-only
3.9.4: Cannot mount root (just tested latest if it was fixed somehow)

I have these kernels in case you want some to boot:
$this->bbcode_second_pass_code('', 'linux-kirkwood-3.5.4-0-arm.pkg.tar.xz
linux-kirkwood-3.6.10-1-arm.pkg.tar.xz
linux-kirkwood-3.6.11-0-arm.pkg.tar.xz
linux-kirkwood-3.6.4-1-arm.pkg.tar.xz
linux-kirkwood-3.7.3-0-arm.pkg.tar.xz
linux-kirkwood-3.7.6-0-arm.pkg.tar.xz
linux-kirkwood-3.7.7-0-arm.pkg.tar.xz
linux-kirkwood-3.7.8-0-arm.pkg.tar.xz
linux-kirkwood-3.8.10-1-arm.pkg.tar.xz
linux-kirkwood-3.8.2-0-arm.pkg.tar.xz
linux-kirkwood-3.8.3-0-arm.pkg.tar.xz
linux-kirkwood-3.8.5-0-arm.pkg.tar.xz
linux-kirkwood-3.8.8-2-arm.pkg.tar.xz
linux-kirkwood-3.9.2-1-arm.pkg.tar.xz
linux-kirkwood-3.9.3-2-arm.pkg.tar.xz
linux-kirkwood-3.9.4-1-arm.pkg.tar.xz')

Thanks for your support.

Edit: and no, write lock isn't set First thing I checked.

[moonman]

There's some other bug involved because it seems esata sheevaplug is the only one that has write_protect gpio initialized. I may have to dig a bit deeper.

[moonman]

I've got a couple of ideas. I'm building the kernels (2) and will post them tomorrow for testing.

[moonman]

Write Protect GPIO disabled on esata sheeva:
$this->bbcode_second_pass_code('', 'http://dl.dropboxusercontent.com/u/15043728/ArchLinuxArm/kernels/testing/esata_sheeva_disable_wp_gpio/linux-kirkwood-3.9.4-2-arm.pkg.tar.xz')

Globally disabled write protect for mmc/sd:
$this->bbcode_second_pass_code('', 'http://dl.dropboxusercontent.com/u/15043728/ArchLinuxArm/kernels/testing/global_disable_sd_wp/linux-kirkwood-3.9.4-3-arm.pkg.tar.xz')

Test and report please. If the first one works, no need to test the second one.

[replabrobin]

Moonman, results for sheeva (basically both 3.9.4-2 & 3.9.4-3 fail). The current 3.9.4-1 is fine.

3.9.4-2 FAILS

$this->bbcode_second_pass_code('', 'U-Boot 2011.12 (Mar 11 2012 - 18:59:46)
Marvell-Sheevaplug - eSATA - SD/MMC

SoC: Kirkwood 88F6281_A1
DRAM: 512 MiB
WARNING: Caches not enabled
NAND: 512 MiB
In: serial
Out: serial
Err: serial
Net: egiga0 [PRIME]
88E1116 Initialized on egiga0
Hit any key to stop autoboot: 0
Marvell>> setenv bootargs $(bootargs_console)
Marvell>> mmc init
SDHC found. Card desciption is:
Manufacturer: 0x02, OEM "TM"
Product name: "SD08G", revision 3.8
Serial number: 3225807631
Manufacturing date: 10/2008
CRC: 0x00, b0 = 0
mmc1 is available
Marvell>> ext2load mmc 0:1 0x800000 uIm3942
Loading file "uIm3942" from mmc device 0:1 (xxa1)
3314776 bytes read
Marvell>> bootm 0x00800000
## Booting kernel from Legacy Image at 00800000 ...
Image Name: Linux-3.9.4-2-ARCH
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 3314712 Bytes = 3.2 MiB
Load Address: 00008000
Entry Point: 00008000
Verifying Checksum ... OK
Loading Kernel Image ... OK
OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.
[ 14.896984] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
[ 14.905314] [<c000dab8>] (unwind_backtrace+0x0/0xe0) from [<c046dcc4>] (panic+0x80/0x1dc)
[ 14.913547] [<c046dcc4>] (panic+0x80/0x1dc) from [<c060fd20>] (mount_block_root+0x260/0x2b4)
[ 14.922032] [<c060fd20>] (mount_block_root+0x260/0x2b4) from [<c0610030>] (prepare_namespace+0x15c/0x1bc)
[ 14.931650] [<c0610030>] (prepare_namespace+0x15c/0x1bc) from [<c060f970>] (kernel_init_freeable+0x1bc/0x204)
[ 14.941617] [<c060f970>] (kernel_init_freeable+0x1bc/0x204) from [<c046bfe8>] (kernel_init+0x8/0xe4)
[ 14.950798] [<c046bfe8>] (kernel_init+0x8/0xe4) from [<c0008d50>] (ret_from_fork+0x14/0x24)')


3.9.4-3 FAILS

$this->bbcode_second_pass_code('', 'U-Boot 2011.12 (Mar 11 2012 - 18:59:46)
Marvell-Sheevaplug - eSATA - SD/MMC

SoC: Kirkwood 88F6281_A1
DRAM: 512 MiB
WARNING: Caches not enabled
NAND: 512 MiB
In: serial
Out: serial
Err: serial
Net: egiga0 [PRIME]
88E1116 Initialized on egiga0
Hit any key to stop autoboot: 0
Marvell>> setenv bootargs $(bootargs_console)
Marvell>> mmc init
SDHC found. Card desciption is:
Manufacturer: 0x02, OEM "TM"
Product name: "SD08G", revision 3.8
Serial number: 3225807631
Manufacturing date: 10/2008
CRC: 0x00, b0 = 0
mmc1 is available
Marvell>> ext2load mmc 0:1 0x800000 uIm3943
Loading file "uIm3943" from mmc device 0:1 (xxa1)
3314824 bytes read
Marvell>> bootm 0x00800000
## Booting kernel from Legacy Image at 00800000 ...
Image Name: Linux-3.9.4-3-ARCH
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 3314760 Bytes = 3.2 MiB
Load Address: 00008000
Entry Point: 00008000
Verifying Checksum ... OK
Loading Kernel Image ... OK
OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.
[ 6.774173] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
[ 6.782503] [<c000dab8>] (unwind_backtrace+0x0/0xe0) from [<c046dcc4>] (panic+0x80/0x1dc)
[ 6.790736] [<c046dcc4>] (panic+0x80/0x1dc) from [<c060fd20>] (mount_block_root+0x260/0x2b4)
[ 6.799222] [<c060fd20>] (mount_block_root+0x260/0x2b4) from [<c0610030>] (prepare_namespace+0x15c/0x1bc)
[ 6.808839] [<c0610030>] (prepare_namespace+0x15c/0x1bc) from [<c060f970>] (kernel_init_freeable+0x1bc/0x204)
[ 6.818806] [<c060f970>] (kernel_init_freeable+0x1bc/0x204) from [<c046bfe8>] (kernel_init+0x8/0xe4)
[ 6.827986] [<c046bfe8>] (kernel_init+0x8/0xe4) from [<c0008d50>] (ret_from_fork+0x14/0x24)')

[ThisIsAandomString]

Will test as I get back from work, in a few hours.

[ThisIsAandomString]

Hi again. 3.9.4-2 works for me.

Maybe we have different eSata Sheevaplug revisions?

Relevant part of the dmesg:

$this->bbcode_second_pass_code('', '[ 16.350857] sdhci: Secure Digital Host Controller Interface driver
[ 16.350866] sdhci: Copyright(c) Pierre Ossman
[ 16.350936] mvsdio mvsdio: no pins associated
[ 16.378460] ata1: SATA link down (SStatus 0 SControl F300)
[ 16.388455] mmc0: mvsdio driver initialized, using GPIO 47 for card detection
[ 16.388519] sdhci-pltfm: SDHCI platform and OF driver helper
[ 16.388576] leds-gpio leds-gpio: pins are not configured from the driver
[ 16.390149] usbcore: registered new interface driver r8712u
[ 16.390529] gre: GRE over IPv4 demultiplexor driver
[ 16.390662] TCP: cubic registered
[ 16.390674] NET: Registered protocol family 17
[ 16.390789] Key type dns_resolver registered
[ 16.393398] registered taskstats version 1
[ 16.394410] console [netcon0] enabled
[ 16.394419] netconsole: network logging started
[ 16.394513] rtc-mv rtc-mv: setting system clock to 2013-05-30 20:55:33 UTC (1369947333)
[ 16.394842] Waiting 1sec before mounting root device...
[ 16.426384] mmc0: host does not support reading read-only switch. assuming write-enable.
[ 16.430330] mmc0: new high speed SDHC card at address b368
[ 16.430639] mmcblk0: mmc0:b368 SDC 30.2 GiB
[ 16.432707] mmcblk0: p1 p2 p3
')

[replabrobin]

Well I don't actually have eSata, mine is the original sheevaplug. So it's the sheevaplug that fails with missing mmc devices.

[moonman]

That's what I was afraid of - fixed one broke the other. I'll compile another kernel tonight when i get home from work.
@replabrobin Is u-boot the same for both devices then? I see yours says sata and you said it isn't the sata version that you have.