[U-Boot] Regression in 2019.01-rc2 for eMMC boot on mxs board?

[U-Boot] Regression in 2019.01-rc2 for eMMC boot on mxs board?

[Michael Heimpold]

Hi,

I'm currently testing latest git tree U-Boot on an I2SE Duckbill 2 device and 
it seems that I've found a regression compared to v2018.11:

Such a Duckbill device is equipped with a NXP i.MX28 CPU and boots from eMMC.
The eMMC has a typical DOS MBR with a special boot loader partition where
U-Boot is stored in. Then U-Boot loads kernel and device tree from an ext 
partition.
When I run latest git U-Boot (see following trace for commit hash), then the
normal boot procedure fails (fallbacks to netboot, fails again) and gives me a 
prompt on Debug UART, see following trace.
The funny part is, that when typing "mmc part" command to list partitions
on eMMC and then calling "boot" again, all works and Linux boot succceeds.

Any quick ideas? Or should I start bi-secting tomorrow...

--snip--
HTLLCLLC

U-Boot 2019.01-rc2-00199-g1cd7b8f120 (Jan 04 2019 - 23:28:55 +0100)

CPU:   Freescale i.MX28 rev1.2 at 454 MHz
BOOT:  SSP SD/MMC #0, 3V3
DRAM:  128 MiB
MMC:   MXS MMC: 0
Loading Environment from MMC... MMC0: Command 8 timeout (status 0xe0384020)
MMC0: Command 55 timeout (status 0xe0284020)
OK
In:    serial
Out:   serial
Err:   serial
Board: I2SE Duckbill 2
Net:   FEC0 [PRIME]
Autobooting in 1 seconds, press <c> to stop
MMC0: Command 8 timeout (status 0xe0384020)
MMC0: Command 55 timeout (status 0xe0284020)
switch to partitions #0, OK
mmc0(part 0) is current device
MMC0: Command 8 timeout (status 0xe0384020)
MMC0: Command 55 timeout (status 0xe0284020)
** No partition table - mmc 0 **
Booting from net ...
BOOTP broadcast 1
DHCP client bound to address 192.168.8.127 (6 ms)
Using FEC0 device
TFTP from server 192.168.8.253; our IP address is 192.168.8.127
Filename 'zImage'.
Load address: 0x42000000
Loading: *
TFTP error: 'File not found' (1)
Not retrying...
BOOTP broadcast 1
DHCP client bound to address 192.168.8.127 (6 ms)
Using FEC0 device
TFTP from server 192.168.8.253; our IP address is 192.168.8.127
Filename 'imx28-duckbill-2.dtb'.
Load address: 0x41000000
Loading: *
TFTP error: 'File not found' (1)
Not retrying...
Bad Linux ARM zImage magic!
=> mmc part

Partition Map for MMC device 0  --   Partition Type: DOS

Part    Start Sector    Num Sectors     UUID            Type
  1     2048            16384           5452574f-01     53 Boot
  2     20480           3731456         5452574f-02     83
=> boot
MMC0: Command 8 timeout (status 0xe0384020)
MMC0: Command 55 timeout (status 0xe0284020)
switch to partitions #0, OK
mmc0(part 0) is current device
MMC0: Command 8 timeout (status 0xe0384020)
MMC0: Command 55 timeout (status 0xe0284020)
3437936 bytes read in 1019 ms (3.2 MiB/s)
Booting from mmc ...
19479 bytes read in 111 ms (170.9 KiB/s)
Kernel image @ 0x42000000 [ 0x000000 - 0x347570 ]
## Flattened Device Tree blob at 41000000
   Booting using the fdt blob at 0x41000000
   Loading Device Tree to 47b80000, end 47b87c16 ... OK

Starting kernel ...

[    0.000000] Booting Linux on physical CPU 0x0
--snap--

Regards,
Michael

[U-Boot] Regression in 2019.01-rc2 for eMMC boot on mxs board?

[Michael Heimpold]

Hi,

this morning I bisected the issue and found the following commit is causing 
it:

d0851c8937067ad396f2bdafc46d0326bf3317db is the first bad commit
commit d0851c8937067ad396f2bdafc46d0326bf3317db
Author: Bin Meng <bmeng.cn@gmail.com>
Date:   Mon Oct 15 02:21:07 2018 -0700

    blk: Call part_init() in the post_probe() method
    
    part_init() is currently called in every DM BLK driver, either
    in its bind() or probe() method. However we can use the BLK
    uclass driver's post_probe() method to do it automatically.
    
    Update all DM BLK drivers to adopt this change.
    
    Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
    Reviewed-by: Simon Glass <sjg@chromium.org>

Just reverting the commit does not apply cleanly, so I tried to manually 
revert the following change:

--- a/drivers/mmc/mmc.c
+++ b/drivers/mmc/mmc.c
@@ -2444,9 +2444,6 @@ static int mmc_startup(struct mmc *mmc)
 	bdesc->product[0] = 0;
 	bdesc->revision[0] = 0;
 #endif
-#if !defined(CONFIG_SPL_BUILD) || defined(CONFIG_SPL_LIBDISK_SUPPORT)
-	part_init(bdesc);
-#endif
.
 	return 0;
 }

After re-adding these lines, all works as expected again and my boot issue is 
gone.

Regards,
mhei

Am Freitag, 4. Januar 2019, 23:48:15 CET schrieb Michael Heimpold:
> Hi,
> 
> I'm currently testing latest git tree U-Boot on an I2SE Duckbill 2 device
> and it seems that I've found a regression compared to v2018.11:
> 
> Such a Duckbill device is equipped with a NXP i.MX28 CPU and boots from
> eMMC. The eMMC has a typical DOS MBR with a special boot loader partition
> where U-Boot is stored in. Then U-Boot loads kernel and device tree from an
> ext partition.
> When I run latest git U-Boot (see following trace for commit hash), then the
> normal boot procedure fails (fallbacks to netboot, fails again) and gives
> me a prompt on Debug UART, see following trace.
> The funny part is, that when typing "mmc part" command to list partitions
> on eMMC and then calling "boot" again, all works and Linux boot succceeds.
> 
> Any quick ideas? Or should I start bi-secting tomorrow...
> 
> --snip--
> HTLLCLLC
> 
> U-Boot 2019.01-rc2-00199-g1cd7b8f120 (Jan 04 2019 - 23:28:55 +0100)
> 
> CPU:   Freescale i.MX28 rev1.2 at 454 MHz
> BOOT:  SSP SD/MMC #0, 3V3
> DRAM:  128 MiB
> MMC:   MXS MMC: 0
> Loading Environment from MMC... MMC0: Command 8 timeout (status 0xe0384020)
> MMC0: Command 55 timeout (status 0xe0284020)
> OK
> In:    serial
> Out:   serial
> Err:   serial
> Board: I2SE Duckbill 2
> Net:   FEC0 [PRIME]
> Autobooting in 1 seconds, press <c> to stop
> MMC0: Command 8 timeout (status 0xe0384020)
> MMC0: Command 55 timeout (status 0xe0284020)
> switch to partitions #0, OK
> mmc0(part 0) is current device
> MMC0: Command 8 timeout (status 0xe0384020)
> MMC0: Command 55 timeout (status 0xe0284020)
> ** No partition table - mmc 0 **
> Booting from net ...
> BOOTP broadcast 1
> DHCP client bound to address 192.168.8.127 (6 ms)
> Using FEC0 device
> TFTP from server 192.168.8.253; our IP address is 192.168.8.127
> Filename 'zImage'.
> Load address: 0x42000000
> Loading: *
> TFTP error: 'File not found' (1)
> Not retrying...
> BOOTP broadcast 1
> DHCP client bound to address 192.168.8.127 (6 ms)
> Using FEC0 device
> TFTP from server 192.168.8.253; our IP address is 192.168.8.127
> Filename 'imx28-duckbill-2.dtb'.
> Load address: 0x41000000
> Loading: *
> TFTP error: 'File not found' (1)
> Not retrying...
> Bad Linux ARM zImage magic!
> => mmc part
> 
> Partition Map for MMC device 0  --   Partition Type: DOS
> 
> Part    Start Sector    Num Sectors     UUID            Type
>   1     2048            16384           5452574f-01     53 Boot
>   2     20480           3731456         5452574f-02     83
> => boot
> MMC0: Command 8 timeout (status 0xe0384020)
> MMC0: Command 55 timeout (status 0xe0284020)
> switch to partitions #0, OK
> mmc0(part 0) is current device
> MMC0: Command 8 timeout (status 0xe0384020)
> MMC0: Command 55 timeout (status 0xe0284020)
> 3437936 bytes read in 1019 ms (3.2 MiB/s)
> Booting from mmc ...
> 19479 bytes read in 111 ms (170.9 KiB/s)
> Kernel image @ 0x42000000 [ 0x000000 - 0x347570 ]
> ## Flattened Device Tree blob at 41000000
>    Booting using the fdt blob at 0x41000000
>    Loading Device Tree to 47b80000, end 47b87c16 ... OK
> 
> Starting kernel ...
> 
> [    0.000000] Booting Linux on physical CPU 0x0
> --snap--
> 
> Regards,
> Michael
> 
> 
> _______________________________________________
> U-Boot mailing list
> U-Boot at lists.denx.de
> https://lists.denx.de/listinfo/u-boot

[U-Boot] Regression in 2019.01-rc2 for eMMC boot on mxs board?

[Simon Glass]

On Sat, 5 Jan 2019 at 01:39, Michael Heimpold <mhei@heimpold.de> wrote:
>
> Hi,
>
> this morning I bisected the issue and found the following commit is causing
> it:
>
> d0851c8937067ad396f2bdafc46d0326bf3317db is the first bad commit
> commit d0851c8937067ad396f2bdafc46d0326bf3317db
> Author: Bin Meng <bmeng.cn@gmail.com>
> Date:   Mon Oct 15 02:21:07 2018 -0700
>
>     blk: Call part_init() in the post_probe() method

Yes, I believe these is a patch to fix this.

Regards,
Simon

[U-Boot] Regression in 2019.01-rc2 for eMMC boot on mxs board?

[Petr Štetiar]

Simon Glass <sjg@chromium.org> [2019-01-07 17:38:31]:

> Yes, I believe these is a patch to fix this.

https://patchwork.ozlabs.org/patch/1014381/

-- ynezz