[PATCH] ARM: zii-vf610-dev: fix boot sd

[PATCH] ARM: zii-vf610-dev: fix boot sd

[Vivien Didelot]

The 'sd' boot is currently broken on zii-vf610-dev because there are no
mci devices (anymore?), only mmc. There are also no need to check
for the device number since both Rev B and Rev C report their SD card
slot on mmc1.

Note that mmc1.0 is already mounted on /mnt/mmc1.0 once mmc1.probe=1
is typed. It would've been more appropriate to use "ln /mnt/mmc1.0
/mnt/sd", but ln isn't useable with automount since the mountpoint
needs to be already created and ln has no force option. There are
no complains with having two mountpoints for a device, so just blindly
mount the device again.

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
---
 .../defaultenv-zii-vf610-dev/init/automount-sd       | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)

diff --git a/arch/arm/boards/zii-vf610-dev/defaultenv-zii-vf610-dev/init/automount-sd b/arch/arm/boards/zii-vf610-dev/defaultenv-zii-vf610-dev/init/automount-sd
index f44dab34e..442c8a61c 100644
--- a/arch/arm/boards/zii-vf610-dev/defaultenv-zii-vf610-dev/init/automount-sd
+++ b/arch/arm/boards/zii-vf610-dev/defaultenv-zii-vf610-dev/init/automount-sd
@@ -1,13 +1,3 @@
 #!/bin/sh
 
-if [ x${global.hostname} = xdev-rev-b -o x${global.hostname} = xdev-rev-c ];
-then
-    global sd=0
-else
-    global sd=1
-fi
-
-mkdir -p /mnt/sd
-automount /mnt/sd 'mci${global.sd}.probe=1 && mount /dev/disk${global.sd}.0 /mnt/sd'
-
-exit 0
+automount -d /mnt/sd 'mmc1.probe=1 && mount /dev/mmc1.0 /mnt/sd'
-- 
2.20.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] ARM: zii-vf610-dev: fix boot sd

[Andrey Smirnov]

On Fri, Jan 18, 2019 at 4:51 PM Vivien Didelot <vivien.didelot@gmail.com> wrote:
>
> The 'sd' boot is currently broken on zii-vf610-dev because there are no
> mci devices (anymore?), only mmc.

Yeah, I forgot to update this part in 5f99a8d403 "dts: VFxxx: Add
aliases for ESDHC controllers", sorry about that!

>  There are also no need to check
> for the device number since both Rev B and Rev C report their SD card
> slot on mmc1.
>

I think you might be misreading the if condition in that check. What
it is meant to test is:

if ((board == VF610 Dev Rev C) || (board == VF610 Dev Rev B)) {
     SDHCI = SDHCI1
} else {
     SDHCI = SDHCI2
}

IOW, it is meant to configure development boards to boot from SD card
(attached to SDHCI1 on both revisions) and from eMMC on other boards
supported by that build (CFU1, SPU3, SCU4).

> Note that mmc1.0 is already mounted on /mnt/mmc1.0 once mmc1.probe=1
> is typed. It would've been more appropriate to use "ln /mnt/mmc1.0
> /mnt/sd", but ln isn't useable with automount since the mountpoint
> needs to be already created and ln has no force option. There are
> no complains with having two mountpoints for a device, so just blindly
> mount the device again.
>

I think we can still take advantage of default automount path
(/mnt/mmc1.0), by moving the if check to "boot/sd" and using
"/mnt/mmc${global.sd}.0/" as a prefix for Linux image and DTB blob.
This way there would be no need to create a symlink or a second mount
point and "init/automount-sd" could be dropped completely.

Thanks,
Andrey Smirnov

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] ARM: zii-vf610-dev: fix boot sd

[Vivien Didelot]

Hi Andrey,

On Sat, 19 Jan 2019 12:39:45 -0800, Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
> On Fri, Jan 18, 2019 at 4:51 PM Vivien Didelot <vivien.didelot@gmail.com> wrote:
> >
> > The 'sd' boot is currently broken on zii-vf610-dev because there are no
> > mci devices (anymore?), only mmc.
> 
> Yeah, I forgot to update this part in 5f99a8d403 "dts: VFxxx: Add
> aliases for ESDHC controllers", sorry about that!

No worries, I'll add a Fixes tag in v2, if Barebox likes that ;-)

> >  There are also no need to check
> > for the device number since both Rev B and Rev C report their SD card
> > slot on mmc1.
> >
> 
> I think you might be misreading the if condition in that check. What
> it is meant to test is:
> 
> if ((board == VF610 Dev Rev C) || (board == VF610 Dev Rev B)) {
>      SDHCI = SDHCI1
> } else {
>      SDHCI = SDHCI2
> }
> 
> IOW, it is meant to configure development boards to boot from SD card
> (attached to SDHCI1 on both revisions) and from eMMC on other boards
> supported by that build (CFU1, SPU3, SCU4).

OK I got confused because I didn't know that zii-vf610-dev was used by
other platforms, I'll keep the check.

> > Note that mmc1.0 is already mounted on /mnt/mmc1.0 once mmc1.probe=1
> > is typed. It would've been more appropriate to use "ln /mnt/mmc1.0
> > /mnt/sd", but ln isn't useable with automount since the mountpoint
> > needs to be already created and ln has no force option. There are
> > no complains with having two mountpoints for a device, so just blindly
> > mount the device again.
> >
> 
> I think we can still take advantage of default automount path
> (/mnt/mmc1.0), by moving the if check to "boot/sd" and using
> "/mnt/mmc${global.sd}.0/" as a prefix for Linux image and DTB blob.
> This way there would be no need to create a symlink or a second mount
> point and "init/automount-sd" could be dropped completely.

Will do, thank you.


	Vivien

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox