jan damborsky wrote:
> I would agree that it doesn't seems to be swap related issue, since it is
> reproducible even if swap is not involved in transfer process.
> Following steps were used in order to hit the problem:
>
> [1] boot into text (CLI) mode with kmdb enabled
> [2] Verify no swap is set and slice 1 is not in use
> [3] create ZFS pool on s0
> [4] create one ZFS filesystem with mountpoint set to legacy
> [5] create /mnt/altroot directory and mount ZFS filesystem on it
> [6] run transfer module binary test application - "transfertest"
>
> Kernel panics when transfer test reaches 39.
>
>
Jan - thanks for the analysis.
> Jan
>
>
> Sarah Jelinek wrote:
>
>> The only major thing that changed in the installer was the change to
>> make swap its own slice, rather than a zvol in the zpool. I will run
>> Bart's latest image and do some debugging. If we had an issue with swap
>> and the root pool overlapping I would think what we would see would be
>> data corruption. And possibly a panic but a panic in ZFS or tmpfs, not
>> hsfs. There is no hsfs on the target disk.
>>
>> sarah
>> ***
>>
>> Moinak Ghosh wrote:
>>
>>
>>> Sanjay Nadkarni wrote:
>>>
>>>
>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> Subject:
>>>> Re: [Fwd: Re: SYNCHED AGAIN: Re: Updated INSTALL package available on
>>>> indiana-build.central]
>>>> From:
>>>> jan damborsky <Jan.Damborsky at Sun.COM>
>>>> Date:
>>>> Thu, 25 Oct 2007 15:35:38 +0200
>>>> To:
>>>> Moinak Ghosh <Moinak.Ghosh at Sun.COM>
>>>>
>>>> To:
>>>> Moinak Ghosh <Moinak.Ghosh at Sun.COM>
>>>> CC:
>>>> Sanjay Nadkarni <Sanjay.Nadkarni at sun.com>, Alok Aggarwal
>>>> <Alok.Aggarwal at Sun.COM>
>>>>
>>>>
>>>> Moinak Ghosh wrote:
>>>>
>>>>
>>>>
>>>>> jan damborsky wrote:
>>>>>
>>>>>
>>>>>
>>>>>> Hi Sanjay,
>>>>>>
>>>>>> I have tried in VMware as well as on W2100z and install crashes
>>>>>> when progress bar reaches 36%. This is a hard crash, system is rebooted
>>>>>> immediately.
>>>>>>
>>>>>> I would like to investigate further, but to tell the truth I don't
>>>>>> know where
>>>>>> to start with debugging since as you pointed out, I can't access the
>>>>>> console
>>>>>> and kmdb hangs if crash occurs when X are running.
>>>>>>
>>>>>> Please let me know, if there might be any suggestions, how to
>>>>>> proceed with
>>>>>> investigation.
>>>>>>
>>>>>>
>>>>>>
>>>>> I am still downloading the image.
>>>>> Since it happens while the transfer module is active, it might be
>>>>> possible to
>>>>> reproduce it via CLI. The transfer module can be rebuilt with make
>>>>> test
>>>>> where it will generate a test binary that can perform the transfer
>>>>> to the root
>>>>> pool mounted on /mnt/altroot.
>>>>>
>>>>>
>>>>>
>>>> I have tried with CLI interface and the excerpt from stack trace after
>>>> crash looks like:
>>>>
>>>> panic[cpu0]/thread=...: assertion failed: len > 0, file:
>>>> ../../common/fs/hsfs/hsfs_vnops.c, line: 1202
>>>>
>>>> genunix:assfail
>>>> hsfs:hsfs_getapage
>>>> genunix:pvn_getpages
>>>> hsfs:hsfsgetpage
>>>> [..]
>>>> lofi:lofi_mapped_rdwr
>>>> lofi:lofi_strategy_task
>>>> genunix:taskq_thread
>>>> unix:thread_start
>>>> [..]
>>>>
>>>> It seems the issue is always reproducible. Please let me know if you
>>>> would need
>>>> more information.
>>>>
>>>>
>>>>
>>> It appears that hsfs_getapage is called with an offset that is at the
>>> end of the file - this should never happen This appears to be the last
>>> segment processing problem in clofi that myself and Alok have been
>>> discussing yesterday. Clofi does not yet handle decompressing the
>>> very last segment properly. We have discussed a fix yesterday. We
>>> need to try that out.
>>>
>>> In addition this ISO is using debug hsfs bits.
>>>
>>> Regards,
>>> Moinak.
>>>
>>>
>>>
>>>
>>>> Thank you,
>>>> Jan
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>> Regards,
>>>>> Moinak.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>> Thank you,
>>>>>> Jan
>>>>>>
>>>>>>
>>>>>> Sanjay Nadkarni wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>> Could either of you guys take a look at this ? I am heading to bed.
>>>>>>> Moinak - any hints on how to debug a system without a console ?
>>>>>>>
>>>>>>> -Sanjay
>>>>>>>
>>>>>>>
>>>>>>> ------------------------------------------------------------------------
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Subject:
>>>>>>> Re: SYNCHED AGAIN: Re: Updated INSTALL package available on
>>>>>>> indiana-build.central
>>>>>>> From:
>>>>>>> Bart Smaalders <bart.smaalders at sun.com>
>>>>>>> Date:
>>>>>>> Wed, 24 Oct 2007 21:26:26 -0700
>>>>>>> To:
>>>>>>> "indiana-eng at sun.com" <indiana-eng at sun.com>
>>>>>>>
>>>>>>> To:
>>>>>>> "indiana-eng at sun.com" <indiana-eng at sun.com>
>>>>>>>
>>>>>>>
>>>>>>> Bart Smaalders wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> Joseph J VLcek wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>> Updated INSTALL package are available on indiana-build.central at:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> /net/indiana-build.central/export/slim_stage/INSTALL/latest
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> New functionality:
>>>>>>>>>
>>>>>>>>> - Add new lower limit 4gb->mb conversion
>>>>>>>>> - Fix swap sizing error.
>>>>>>>>> - Misc nits cleanup.
>>>>>>>>> - Modify slice setup to ensure swap on slice not zvol.
>>>>>>>>> - Add swap device to /etc/vfstab
>>>>>>>>> - Include sbin/enable_nwam and var/svc/profile/network_nwam.xml
>>>>>>>>> to ensure nwam is enabled at reboot.
>>>>>>>>> - TI_ATTR_SLICE_WDISK_FL renamed to TI_ATTR_SLICE_DEFAULT_LAYOUT
>>>>>>>>> - s0&s1 are created when default VTOC configuration is required
>>>>>>>>> - set up swap space on s1 instead of ZFS volume
>>>>>>>>> - rename to Indiana Developer
>>>>>>>>>
>>>>>>>>> I have done a basic sanity check.
>>>>>>>>>
>>>>>>>>> Joe
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>> I've reloaded slim_install and SUNWos86r into cyber:30000
>>>>>>>> again to pick up David's fix for the pci_pci bridge aliases
>>>>>>>> problem (Thanks Vikram & Jan!)
>>>>>>>>
>>>>>>>> This should address some of the lingering device problems
>>>>>>>> we've been seing.
>>>>>>>>
>>>>>>>> - Bart
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>> There is a distro built from these bits at
>>>>>>>
>>>>>>> /home/barts/isos/working_version.iso
>>>>>>>
>>>>>>> It crashes (panics) during install on my macbook.
>>>>>>>
>>>>>>> - Bart
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> _______________________________________________
>>>> caiman-discuss mailing list
>>>> caiman-discuss at opensolaris.org
>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>>
>>>>
>>>>
>>>>
>>> _______________________________________________
>>> caiman-discuss mailing list
>>> caiman-discuss at opensolaris.org
>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>
>>>
>>>
>>>
>> _______________________________________________
>> caiman-discuss mailing list
>> caiman-discuss at opensolaris.org
>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>
>>
>
> _______________________________________________
> caiman-discuss mailing list
> caiman-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>
