| Message ID | 33a698b367be37f4ea6015162401f55161c4279e.1427142428.git.mschiffer@universe-factory.net |
|---|---|
| State | Changes Requested |
| Headers | show |
I tested your code with my old WNDR3700 (v1), WNDR3700v2 and WNDR3800. All three routers got detected ok. My v2 is "old" and its art contents stop at byte 0x28, so it uses your "\xff" branch. Just for reference, an old bug with discussion (from you?) about this issue: https://dev.openwrt.org/ticket/18992 Ps. I think that you mixed the explanation regarding the old logic: The first W gets cut away when the ID is shorter, not when it is longer. 3800 would have "...+128WND..." and correct ID, but new 3700v2 would have "...+64WNDR..." at the same place and the detected ID would miss the first W.
On 03/25/2015 05:46 PM, Hannu Nyman wrote: > I tested your code with my old WNDR3700 (v1), WNDR3700v2 and WNDR3800. > All three routers got detected ok. > > My v2 is "old" and its art contents stop at byte 0x28, so it uses your > "\xff" branch. Ah, can you send my a dump of your ART? Then I can hopefully simplify the patch a bit. > > Just for reference, an old bug with discussion (from you?) about this > issue: > https://dev.openwrt.org/ticket/18992 Yes, that's from me. > > > Ps. > I think that you mixed the explanation regarding the old logic: The > first W gets cut away when the ID is shorter, not when it is longer. > 3800 would have "...+128WND..." and correct ID, but new 3700v2 would > have "...+64WNDR..." at the same place and the detected ID would miss > the first W. Yes, you're right. If I send a v2 of this, I'll fix the explanation.
diff --git a/target/linux/ar71xx/base-files/lib/ar71xx.sh b/target/linux/ar71xx/base-files/lib/ar71xx.sh index d2fe6ef..4781dd7 100755 --- a/target/linux/ar71xx/base-files/lib/ar71xx.sh +++ b/target/linux/ar71xx/base-files/lib/ar71xx.sh @@ -37,16 +37,28 @@ wndr3700_board_detect() { machine="NETGEAR WNDR3700" ;; "33373031") - local model - model=$(ar71xx_get_mtd_offset_size_format art 56 10 %c) - if [ -z "$model" ] || [ "$model" = $'\xff\xff\xff\xff\xff\xff\xff\xff\xff\xff' ]; then + case "$(ar71xx_get_mtd_offset_size_format art 56 10 %c)" in + $'\xff\xff\xff\xff\xff\xff\xff\xff\xff\xff') machine="NETGEAR WNDR3700v2" - elif [ -z "$model" ] || [ "$model" = $'\xff\xff\xff\xff\xff\xff\xff\xff\xffN' ]; then + ;; + $'\xff\xff\xff\xff\xff\xff\xff\xff\xffN') machine="NETGEAR WNDRMAC" - else - machine="NETGEAR $model" - fi - ;; + ;; + *) + # Use awk to remove everything after the first zero byte + model="$(ar71xx_get_mtd_offset_size_format art 41 32 %c | awk 'BEGIN{FS="[[:cntrl:]]"} {print $1; exit}')" + case $model in + '29763654+16+64'*) + machine="NETGEAR ${model:14}" + ;; + '29763654+16+128'*) + machine="NETGEAR ${model:15}" + ;; + *) + # Unknown ID + machine="NETGEAR $model" + esac + esac esac AR71XX_BOARD_NAME="$name"
There were a few issues with the existing code to detect the model string: * Always using the string starting with byte 56 would cut off the W of WNDR when the ID starts with 29763654+16+128 instead of 29763654+16+64 * The string contained garbage after the zero byte instead of cutting it off after the zero (which wasn't always visible using busybox tools, but could confuse other scripts) Tested on a WNDR3700v1 and WNDR3700v2. Unfortunately I couldn't find any devices which use the $'\xff.....' branch, all newer revisions of the WNDR3700v2 and WNDRMAC devices actually contain their model name in the ART and thus use the * branch as well. The [ -z "$model" ] check was dropped as there is no way to actually hit this unless no ART is found at all. The awk command was carefully crafted to work both with gawk and the (horribly broken) busybox awk. Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net> --- target/linux/ar71xx/base-files/lib/ar71xx.sh | 28 ++++++++++++++++++++-------- 1 file changed, 20 insertions(+), 8 deletions(-)