Bug#683329: flash-kernel: Toshiba AC100 support broken [fix included]
Control: found -1 3.0~rc.2
Control: found -1 3.3+deb7u2
Control: found -1 3.11
Control: tag -1 + patch
Control: severity -1 important
Hi,
Thomas Maass wrote:
> flash-kernel no longer flashs the Toshiba AC100 (Tegra2).
> It shows only Installing version... but no Flashing...
> The last version I remember to work was 3.0rc1.
Actually it was 3.0~rc.1. And the first one which no more worked was
3.0~rc.2.
Ran into that too when switching my AC100 from Ubuntu Precise to
Debian Wheezy -- and it's still present in Debian Jessie/Sid.
Vincent Zweije wrote:
> Trying to flash a new kernel failed - flash-kernel exits early because
> some shell command returned an error code. I have tracked this down to
> faulty boot device detection.
>
> This little loop from /usr/share/flash-kernel/functions around line 468
> does the detection:
>
> for p in "$android_boot_device"*[0-9]; do
> abootimg="$(LC_ALL=C abootimg -i "$p" 2>/dev/null)"
> image_size="$(abootimg_get_image_size "$abootimg")"
> if [ -n "$image_size" ] &&
> [ "$image_size" -gt "$largest_size" ]; then
> part="$p"
> fi
> done
>
> The assignment to abootimg contains a call to abootimg -i "$p". This
> call fails when the glob pattern on the line above returns a block device
> that is not an android boot image. Since the script is executed with -e,
> this aborts the script.
>
> Adding a "|| :" after the command fixes the problem thus:
>
> for p in "$android_boot_device"*[0-9]; do
> abootimg="$(LC_ALL=C abootimg -i "$p" 2>/dev/null || : )"
> image_size="$(abootimg_get_image_size "$abootimg")"
> if [ -n "$image_size" ] &&
> [ "$image_size" -gt "$largest_size" ]; then
> part="$p"
> fi
> done
I came to the same conclusion and patch:
--- functions.orig 2013-10-25 13:58:31.379524139 +0200
+++ functions 2013-10-25 13:40:48.984844024 +0200
@@ -475,7 +475,7 @@
part=""
largest_size="-1"
for p in "$android_boot_device"*[0-9]; do
- abootimg="$(LC_ALL=C abootimg -i "$p" 2>/dev/null)"
+ abootimg="$(LC_ALL=C abootimg -i "$p" 2>/dev/null || true)"
image_size="$(abootimg_get_image_size "$abootimg")"
if [ -n "$image_size" ] &&
[ "$image_size" -gt "$largest_size" ]; then
Tagging the bug report accordingly and raising the severity to
important as this definitely affects all AC100 users.
The probably easiest work-around (i.e. one which doesn't involve
patching) is to use Julian's packages from
http://people.debian.org/~jak/ac100/ which are apt-get-able.
JFTR: Here's a condensed summary of what happens:
# for i in /dev/mmcblk0p*; do echo $i; abootimg -i $i | fgrep 'image size'; abootimg -i $i >/dev/null 2>&1 ; echo $?; done
/dev/mmcblk0p1
* image size = 5242880 bytes (5.00 MB)
0
/dev/mmcblk0p2
* image size = 8388608 bytes (8.00 MB)
0
/dev/mmcblk0p3
/dev/mmcblk0p3: no Android Magic Value
/dev/mmcblk0p3: not a valid Android Boot Image.
1 ← This is where flash-kernel aborts due to "set -e".
/dev/mmcblk0p4
/dev/mmcblk0p4: no Android Magic Value
/dev/mmcblk0p4: not a valid Android Boot Image.
1
/dev/mmcblk0p5
/dev/mmcblk0p5: no Android Magic Value
/dev/mmcblk0p5: not a valid Android Boot Image.
1
/dev/mmcblk0p6
/dev/mmcblk0p6: no Android Magic Value
/dev/mmcblk0p6: not a valid Android Boot Image.
1
/dev/mmcblk0p7
/dev/mmcblk0p7: no Android Magic Value
/dev/mmcblk0p7: not a valid Android Boot Image.
1
#
Regards, Axel
--
,''`. | Axel Beckert <abe@debian.org>, http://people.debian.org/~abe/
: :' : | Debian Developer, ftp.ch.debian.org Admin
`. `' | 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
`- | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5
Reply to: