| Summary: | PTEST image does not install on hdd | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] BSPs | Reporter: | Lucian Musat <georgex.l.musat> | ||||||||
| Component: | bsps-runtime | Assignee: | Leonardo Sandoval Gonzalez <leonardo.sandoval.gonzalez> | ||||||||
| Status: | VERIFIED FIXED | QA Contact: | Lucian Musat <georgex.l.musat> | ||||||||
| Severity: | major | ||||||||||
| Priority: | Medium+ | CC: | alejandro.hernandez, alexandru.c.georgescu, benjamin.esquivel, costin.c.constantin, leonardo.sandoval.gonzalez, yp.kernel.watcher, yp.watcher | ||||||||
| Version: | 1.9 | ||||||||||
| Target Milestone: | 1.9 M3 | ||||||||||
| Hardware: | x86 | ||||||||||
| OS: | x86_64 | ||||||||||
| Whiteboard: | |||||||||||
| OS type for building Yocto: | --- | Type of Regression: | Regression (Used to work) | ||||||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||||||
| Attachments: |
|
||||||||||
|
Description
Lucian Musat
2015-07-08 15:34:05 UTC
I tried to boot this image both with a NUC and via runqemu using the following command: runqemu qemux86-64 ~/Downloads/bzImage-genericx86-64.bin ~/Downloads/core-image-sato-sdk-genericx86-64.ext3 In both cases it booted to the Splash screen and Sato UI, I did not see any kind of failure, is it possibly your image is corrupted? 15f4c8b9f9b6bd8437e373e72623cb10 /home/sgw/Downloads/core-image-sato-sdk-genericx86-64.ext3 e3ead06891e27bbd4566d16dbfe5cf07 /home/sgw/Downloads/core-image-sato-sdk-genericx86-64.hddimg I would assume the same failure would occur with the .ext3 since it's a startup -> cmd prompt issue, but both worked correctly. Please verify your image and re-flash it to the USB Stick or SD card. Possibly try a new USB Stick. Lucian reported that it worked using a fresh image and usb stick, so I would gather a corrupted image along the way. For what is worth I re-tested this and it seems that it works when booting the stick via legacy mode and fails as I reported when booting it in EFI mode. Using our sticks anyway. Re-opening this for further investigation because it blocks our automatic tests in the lab. (In reply to comment #3) > For what is worth I re-tested this and it seems that it works when booting > the stick via legacy mode and fails as I reported when booting it in EFI > mode. Using our sticks anyway. Lucian: Could you provide the steps to boot in both modes (legacy and EFI)? When starting a machine press F10 to enter boot menu, and from there you should have the option to boot the stick normally or efi mode. If not, make sure you have checked "UEFI boot" option in bios. (In reply to comment #6) > When starting a machine press F10 to enter boot menu, and from there you > should have the option to boot the stick normally or efi mode. If not, make > sure you have checked "UEFI boot" option in bios. Hi Lucian, I have tried the kernel/rootfs from 1.9 M1.rc2 [1] using a qemux86-64 machine. As expected, things looks on this machine (the issue is not found). I will then try on a HW (minowboard) and with EFI mode. I let you know my results asap. [1] http://autobuilder.yoctoproject.org/pub/releases/yocto-yocto-1.9_M1.rc2/ptest/machines/genericx86-64/ (In reply to comment #7) > (In reply to comment #6) > > When starting a machine press F10 to enter boot menu, and from there you > > should have the option to boot the stick normally or efi mode. If not, make > > sure you have checked "UEFI boot" option in bios. > > Hi Lucian, > I have tried the kernel/rootfs from 1.9 M1.rc2 [1] using a qemux86-64 > machine. As expected, things looks on this machine (the issue is not found). > I will then try on a HW (minowboard) and with EFI mode. I let you know my > results asap. > > [1] > http://autobuilder.yoctoproject.org/pub/releases/yocto-yocto-1.9_M1.rc2/ > ptest/machines/genericx86-64/ Hi, Did you manage to replicate the issue? (In reply to comment #8) > (In reply to comment #7) > > (In reply to comment #6) > > > When starting a machine press F10 to enter boot menu, and from there you > > > should have the option to boot the stick normally or efi mode. If not, make > > > sure you have checked "UEFI boot" option in bios. > > > > Hi Lucian, > > I have tried the kernel/rootfs from 1.9 M1.rc2 [1] using a qemux86-64 > > machine. As expected, things looks on this machine (the issue is not found). > > I will then try on a HW (minowboard) and with EFI mode. I let you know my > > results asap. > > > > [1] > > http://autobuilder.yoctoproject.org/pub/releases/yocto-yocto-1.9_M1.rc2/ > > ptest/machines/genericx86-64/ > > Hi, > Did you manage to replicate the issue? Alex, Sorry for the delay. I tried on a Minnowboard MAX, using a [2] with this (md5sum) checksum 0db9d820a4d73e082fb28c29280b9741 core-image-sato-sdk-genericx86-64.hddimg There are the test I did 1. To boot in EFI mode, once board is power on, I typed the following Shell> fs0: FS0:\> cd EFI/BOOT FS0:\> bootx64.efi This leads to login prompt, log as root and run the ptest-runner. So far, all tests have passed correctly as seen in the below log: . . PASS: 280 - Start tracing for session valid_filter PASS: 281 - Stop lttng tracing for session valid_filter PASS: 282 - Destroy session valid_filter PASS: 283 - Validate trace filter output PASS: 284 - Create session valid_filter in /tmp/tmp.RF7f0U3uuM PASS: 285 - Enable lttng event with filtering PASS: 286 - Start tracing for session valid_filter PASS: 287 - Stop lttng tracing for session valid_filter PASS: 288 - Destroy session valid_filter PASS: 289 - Validate trace filter output PASS: 290 - Kill session daemon Unless I am not booting in EFI mode, this issue is NOT reproduced on the M1.rc2 ptest sato-sdk image. [2] http://autobuilder.yoctoproject.org/pub/releases/yocto-yocto-1.9_M1.rc2/ptest/machines/genericx86-64/core-image-sato-sdk-genericx86-64.hddimg Did you select install or live boot? The install is the one that fails. We tried again now with a brand new minnowboard max and it still fails for us. We tried on all kinds of machines. We'll try to upload a video of the fail. Created attachment 2610 [details]
the installation process as it is now
Hello, The last attachment contains a small clip with the installation as it is now. Lucian/Costin
The problem is that I am not using the same image.
Can you share the link of the image so I can test the same?
So far, I have used [2], instrument the grub.cfg file to include the install option (I only have the 'boot' option, which I believe this is what you call 'live boot'):
menuentry 'install'{
linux /vmlinuz LABEL=install-efi root=/dev/ram0
initrd /initrd
}
and boot, but this is not booting the same way as seen on the video (no kernel messages and no automatic installation).
Hi, The image we used is the rc2 one: http://autobuilder.yoctoproject.org/pub/releases/yocto-yocto-1.9_M1.rc2/ptest/machines/genericx86-64/core-image-sato-sdk-genericx86-64.hddimg Just write it to an usb drive, stick it in a minnowboard max, and boot it from efi shell. But make sure that you are booting the drive and not the hdd. Usually fs0: is the hdd. I was able to reproduce the issue. This time I used dd instead of mkefidisk.sh, and connect a monitor instead of the serial console. I will work on finding the root cause ASAP. Issue is reproduced using the image below (image produced by local builder, talking the latest master at midnight of the indicated day). This means that the problem is present on master and it is *not* related to ptest. http://yctb01.zpn.intel.com/builds/nightly-x86-64/20150727-1/machines/genericx86-64/core-image-sato-sdk-genericx86-64.hddimg looking into the root cause... Created attachment 2616 [details] kernel log on a non-ptest image Log taken using this image: http://yctb01.zpn.intel.com/builds/nightly-x86-64/20150727-1/machines/genericx86-64/core-image-sato-sdk-genericx86-64.hddimg on a minnowboardmax board. An image can only be installed on a device different that the live one (live OS), so what we are seeing is expected. However, the output message
"No hard drive selected. Installation aborted"
is not clear enough, and this needs a better message or a check before this happens. I will send a patch for the latter.
Patch sent to the OE-Core mailing list. (In reply to comment #18) > An image can only be installed on a device different that the live one (live > OS), so what we are seeing is expected. However, the output message > > "No hard drive selected. Installation aborted" > > is not clear enough, and this needs a better message or a check before this > happens. I will send a patch for the latter. Hello Leonardo, I am not sure I follow what you are saying. We are trying to install from USB stick onto the hard drive. The script clearly shows that it detects the HDD (MODEL=INTEL SSDSC2...) so the source is not the same as the destination. (In reply to comment #20) > (In reply to comment #18) > > An image can only be installed on a device different that the live one (live > > OS), so what we are seeing is expected. However, the output message > > > > "No hard drive selected. Installation aborted" > > > > is not clear enough, and this needs a better message or a check before this > > happens. I will send a patch for the latter. > > Hello Leonardo, > > I am not sure I follow what you are saying. We are trying to install from > USB stick onto the hard drive. The script clearly shows that it detects the > HDD (MODEL=INTEL SSDSC2...) so the source is not the same as the destination. I have reproduced the issue you are seeing (for some reason, I though you were trying to boot & install at the same device). I will send a V2 patch as soon as I find the root cause. A new patch sent to the community (titled '[PATCH] init-install-efi.sh: Avoid /mnt/mtab creation if already present'). Under review now. The patch mentioned by Leo in the immediate previous comment is located at: http://lists.openembedded.org/pipermail/openembedded-core/2015-August/108283.html It looks good to me. Patch integrated into master (5acf99d1afc7efa962e4e698cc09a7f08133a64e) Seems to be installing now. Verified on commit eac61f37e36099f74485dab398b57f3812826d17. |