Testing Env: Host: Ubuntu 10.10 x86_64 Kernel: 2.6.35-28-generic Image location: http://autobuilder.yoctoproject.org/nightly/20110527-1/machines/beagleboard/ Tree/Branch: Poky/1.1_M1 Poky Commit: 4ff7af11ef69849ef9c16f585eae58ac920b222b When boot up beagleboard with minimal or sato image, 'init' will print some messages: #################### ... Starting syslogd/klogd: done * Starting Avahi mDNS/DNS-SD Daemon: avahi-daemon ...done. Starting tcf-agent: OK INIT: Id "S" respawning too fast: disabled for 5 minutes INIT: Id "S" respawning too fast: disabled for 5 minutes INIT: Id "S" respawning too fast: disabled for 5 minutes INIT: Id "S" respawning too fast: disabled for 5 minutes INIT: Id "S" respawning too fast: disabled for 5 minutes INIT: Id "S" respawning too fast: disabled for 5 minutes ################## The message will loop forever and can not login. With 20110520-2 build (git commit: fc55b224caa3eeac4abda099ec9ed505db59fb28), there is no this issue. I diff /etc/inittab with these two build: ############### diff -Nurp 20/etc/inittab 27/etc/inittab --- 20/etc/inittab 2011-05-21 21:39:10.000000000 +0800 +++ 27/etc/inittab 2011-05-29 06:06:03.000000000 +0800 @@ -28,7 +28,7 @@ l5:5:wait:/etc/init.d/rc 5 l6:6:wait:/etc/init.d/rc 6 # Normally not reached, but fallthrough in case of emergency. z6:6:respawn:/sbin/sulogin -S:2345:respawn:/sbin/getty 115200 ttyO2 +S:2345:respawn:/sbin/getty 115200 ttyAMA0 # /sbin/getty invocations for the runlevels. # # The "id" field MUST be the same as the last ############### When I modified ‘ttyAMA0' to 'ttyO2', it works. My u-boot setting: ############### setenv bootcmd 'mmc rescan 0; mmc init; fatload mmc 0:1 0x80300000 uImage; bootm 0x80300000' setenv bootargs 'console=tty0 console=ttyO2,115200n8 root=/dev/mmcblk0p2 rootwait rootfstype=ext3 ro' boot ###############
That looks like a misconfiguration on the autobuilder to me. meta-yocto beagleboard machine still specifies: SERIAL_CONSOLE = "115200 ttyO2" Has anything changed on the autobuilder that might modify SERIAL_CONSOLE?
No, nothing should have changed from my end.
(In reply to comment #1) > That looks like a misconfiguration on the autobuilder to me. meta-yocto > beagleboard machine still specifies: > > SERIAL_CONSOLE = "115200 ttyO2" > > Has anything changed on the autobuilder that might modify SERIAL_CONSOLE? http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta-yocto/conf/machine/beagleboard.conf?h=1.1_M1 On the 1.1_M1 branch, this is still in the meta-yocto beagleboard conf. I'll look for the patch and cherry-pick into branch.
(In reply to comment #3) > (In reply to comment #1) > > That looks like a misconfiguration on the autobuilder to me. meta-yocto > > beagleboard machine still specifies: > > > > SERIAL_CONSOLE = "115200 ttyO2" > > > > Has anything changed on the autobuilder that might modify SERIAL_CONSOLE? > > http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta-yocto/conf/machine/beagleboard.conf?h=1.1_M1 > > On the 1.1_M1 branch, this is still in the meta-yocto beagleboard conf. I'll > look for the patch and cherry-pick into branch. Then it is correct as is - sorry if my wording was confusing. What we would be looking for is something that changes SERIAL_CONSOLE to something other than what I have listed above. Specifically, to something like: SERIAL_CONSOLE = "115200 ttyAMA0" That is what would be causing the breakage.
(In reply to comment #4) > (In reply to comment #3) > > (In reply to comment #1) > > > That looks like a misconfiguration on the autobuilder to me. meta-yocto > > > beagleboard machine still specifies: > > > > > > SERIAL_CONSOLE = "115200 ttyO2" > > > > > > Has anything changed on the autobuilder that might modify SERIAL_CONSOLE? > > > > http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta-yocto/conf/machine/beagleboard.conf?h=1.1_M1 > > > > On the 1.1_M1 branch, this is still in the meta-yocto beagleboard conf. I'll > > look for the patch and cherry-pick into branch. > > Then it is correct as is - sorry if my wording was confusing. What we would be > looking for is something that changes SERIAL_CONSOLE to something other than > what I have listed above. Specifically, to something like: > > SERIAL_CONSOLE = "115200 ttyAMA0" > > That is what would be causing the breakage. I've looked and nothing appears to be doing that.
I built core-image-sato w/ 4ff7af11ef69849ef9c16f585eae58ac920b222b and /etc/inittab in the generated rootfs contains the correct line: S:2345:respawn:/sbin/getty 115200 ttyO2 Checking the root image of: http://autobuilder.yoctoproject.org/nightly/20110527-1/machines/beagleboard/arm/core-image-minimal-beagleboard-20110529133331.rootfs.tar.bz2 and http://autobuilder.yoctoproject.org/nightly/20110527-1/machines/beagleboard/arm/core-image-sato-beagleboard-20110529133331.rootfs.tar.bz2 reveals both have the bad getty line in inittab before initial boot: S:2345:respawn:/sbin/getty 115200 ttyAMA0 A quick git grep reveals: $ git grep AMA0 meta/conf/machine/qemuarm.conf:SERIAL_CONSOLE = "115200 ttyAMA0" This suggests some confusion to me on the part of the autobuilder regarding which machine it is building for since the files do indeed include "beagleboard" in the name, but it is picking up the SERIAL_CONSOLE from qemuarm.conf. Beth, can you have a look at the autobuilder local.conf to see if anything there might give us a hint? Does the build try and share any sstate with qemuarm?
(In reply to comment #6) > I built core-image-sato w/ 4ff7af11ef69849ef9c16f585eae58ac920b222b and > /etc/inittab in the generated rootfs contains the correct line: > > S:2345:respawn:/sbin/getty 115200 ttyO2 > > Checking the root image of: > http://autobuilder.yoctoproject.org/nightly/20110527-1/machines/beagleboard/arm/core-image-minimal-beagleboard-20110529133331.rootfs.tar.bz2 > > and > > http://autobuilder.yoctoproject.org/nightly/20110527-1/machines/beagleboard/arm/core-image-sato-beagleboard-20110529133331.rootfs.tar.bz2 > > reveals both have the bad getty line in inittab before initial boot: > > S:2345:respawn:/sbin/getty 115200 ttyAMA0 > > A quick git grep reveals: > > $ git grep AMA0 > meta/conf/machine/qemuarm.conf:SERIAL_CONSOLE = "115200 ttyAMA0" > > This suggests some confusion to me on the part of the autobuilder regarding > which machine it is building for since the files do indeed include > "beagleboard" in the name, but it is picking up the SERIAL_CONSOLE from > qemuarm.conf. > > Beth, can you have a look at the autobuilder local.conf to see if anything > there might give us a hint? Does the build try and share any sstate with > qemuarm? Yes. It does share sstate with qemuarm. If it can't share sstate, then this may cause some performance issues on the AB.
This looks like the opposite problem to an sstate issue. If sstate basichash was enabled we'd actually not see this problem! The bug looks like sysvinit needs PACKAGE_ARCH = "${MACHINE_ARCH}" adding to it since the sysvinit-inittab package is machine specific (correctly) but the recipe is only being built per arch (roughly speaking). The alternative is to split out the conf file into a separate machine specific recipe.
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=5d9ea7dea48e640b7345b7d664b284f731a4d6f9
*** Bug 1071 has been marked as a duplicate of this bug. ***
Merged to M1 now also
Verified it with 20110602 build Tree/Branch: Poky/1.1_M1 Poky Commit: b4a215bd94cf1508782e49b1a3f8c6b1f7effbb9
Verified it with 20110613 build Image location: http://autobuilder.yoctoproject.org/nightly/20110613-1/ Tree/Branch: Poky/master Commit: e5f3cc34df74c08c8cab3e1cf9207fafaec6cb9a