Unable to boot linux-next build

Unable to boot linux-next build

[Quentin Lambert]

 I have been trying to build and boot the last version available on linux-next.
 During the build I am being prompted with "has no CRC!" warnings for a bunch
 of modules.

 Has a result I get the following lines in the Module.symvers:

 [qlambert@sloth linux-next]$ grep 0x00000 Module.symvers
 0x00000000	phys_base	vmlinux	EXPORT_SYMBOL
 0x00000000	memmove	vmlinux	EXPORT_SYMBOL
 0x00000000	__copy_user_nocache	vmlinux	EXPORT_SYMBOL
 0x00000000	__get_user_4	vmlinux	EXPORT_SYMBOL
 0x00000000	__put_user_4	vmlinux	EXPORT_SYMBOL
 0x00000000	__memcpy	vmlinux	EXPORT_SYMBOL
 0x00000000	__fentry__	vmlinux	EXPORT_SYMBOL
 0x00000000	memset	vmlinux	EXPORT_SYMBOL
 0x00000000	__sw_hweight64	vmlinux	EXPORT_SYMBOL
 0x00000000	__sw_hweight32	vmlinux	EXPORT_SYMBOL
 0x00000000	memcpy_mcsafe_unrolled	vmlinux	EXPORT_SYMBOL_GPL
 0x00000000	memcpy	vmlinux	EXPORT_SYMBOL
 0x00000000	copy_user_enhanced_fast_string	vmlinux	EXPORT_SYMBOL
 0x00000000	clear_page	vmlinux	EXPORT_SYMBOL
 0x00000000	__put_user_2	vmlinux	EXPORT_SYMBOL
 0x00000000	__get_user_2	vmlinux	EXPORT_SYMBOL
 0x00000000	copy_page	vmlinux	EXPORT_SYMBOL
 0x00000000	copy_user_generic_string	vmlinux	EXPORT_SYMBOL
 0x00000000	_copy_to_user	vmlinux	EXPORT_SYMBOL
 0x00000000	___preempt_schedule	vmlinux	EXPORT_SYMBOL
 0x00000000	__memmove	vmlinux	EXPORT_SYMBOL
 0x00000000	empty_zero_page	vmlinux	EXPORT_SYMBOL
 0x00000000	__get_user_8	vmlinux	EXPORT_SYMBOL
 0x00000000	__put_user_8	vmlinux	EXPORT_SYMBOL
 0x00000000	_copy_from_user	vmlinux	EXPORT_SYMBOL
 0x00000000	native_load_gs_index	vmlinux	EXPORT_SYMBOL
 0x00000000	__memset	vmlinux	EXPORT_SYMBOL
 0x00000000	___preempt_schedule_notrace	vmlinux	EXPORT_SYMBOL
 0x00000000	__put_user_1	vmlinux	EXPORT_SYMBOL
 0x00000000	__get_user_1	vmlinux	EXPORT_SYMBOL
 0x00000000	copy_user_generic_unrolled	vmlinux	EXPORT_SYMBOL

 If I ignore these and go on to boot the resulting kernel. I am left with:

 ERROR: Unable to find root device 'UUID=<omitted>'

 I am then dropped to a recovery shell where the keyboard is not answering.

 The UUID is valid and is the same being used when I boot the HEAD of the linux
 tree. In fact, I am sending this message from the same machine, running
 that very kernel.
 Prior to failing to find the root device, the kernel prints a series of
 messages for each of the above symbol:

 scsi_mod: no symbol version for <foo>
 scsi_mod: Unknown symbol <foo> (err -22)

 Therefore, I suspect that the missing CRC messages are linked to the issue.

 I have git bisected this behavior to the patch 784d5699edd. 

 Simply reverting this patch results in different issues.
 The errors I am having due to no CRC are fixed but I receive another error at
 boot time:
 failed command: WRITE LOG DMA EXT

 I mostly asking for help as to what to do next.
 Could someone point me toward documentation that would help me understand and
 fix this issue?

Re: Unable to boot linux-next build

[Quentin Lambert]

On 09/09/2016 10:31, Quentin Lambert wrote:
>   I have been trying to build and boot the last version available on linux-next.
>   During the build I am being prompted with "has no CRC!" warnings for a bunch
>   of modules.
>
>   Has a result I get the following lines in the Module.symvers:
>
>   [qlambert@sloth linux-next]$ grep 0x00000 Module.symvers
>   0x00000000	phys_base	vmlinux	EXPORT_SYMBOL
>   0x00000000	memmove	vmlinux	EXPORT_SYMBOL
>   0x00000000	__copy_user_nocache	vmlinux	EXPORT_SYMBOL
>   0x00000000	__get_user_4	vmlinux	EXPORT_SYMBOL
>   0x00000000	__put_user_4	vmlinux	EXPORT_SYMBOL
>   0x00000000	__memcpy	vmlinux	EXPORT_SYMBOL
>   0x00000000	__fentry__	vmlinux	EXPORT_SYMBOL
>   0x00000000	memset	vmlinux	EXPORT_SYMBOL
>   0x00000000	__sw_hweight64	vmlinux	EXPORT_SYMBOL
>   0x00000000	__sw_hweight32	vmlinux	EXPORT_SYMBOL
>   0x00000000	memcpy_mcsafe_unrolled	vmlinux	EXPORT_SYMBOL_GPL
>   0x00000000	memcpy	vmlinux	EXPORT_SYMBOL
>   0x00000000	copy_user_enhanced_fast_string	vmlinux	EXPORT_SYMBOL
>   0x00000000	clear_page	vmlinux	EXPORT_SYMBOL
>   0x00000000	__put_user_2	vmlinux	EXPORT_SYMBOL
>   0x00000000	__get_user_2	vmlinux	EXPORT_SYMBOL
>   0x00000000	copy_page	vmlinux	EXPORT_SYMBOL
>   0x00000000	copy_user_generic_string	vmlinux	EXPORT_SYMBOL
>   0x00000000	_copy_to_user	vmlinux	EXPORT_SYMBOL
>   0x00000000	___preempt_schedule	vmlinux	EXPORT_SYMBOL
>   0x00000000	__memmove	vmlinux	EXPORT_SYMBOL
>   0x00000000	empty_zero_page	vmlinux	EXPORT_SYMBOL
>   0x00000000	__get_user_8	vmlinux	EXPORT_SYMBOL
>   0x00000000	__put_user_8	vmlinux	EXPORT_SYMBOL
>   0x00000000	_copy_from_user	vmlinux	EXPORT_SYMBOL
>   0x00000000	native_load_gs_index	vmlinux	EXPORT_SYMBOL
>   0x00000000	__memset	vmlinux	EXPORT_SYMBOL
>   0x00000000	___preempt_schedule_notrace	vmlinux	EXPORT_SYMBOL
>   0x00000000	__put_user_1	vmlinux	EXPORT_SYMBOL
>   0x00000000	__get_user_1	vmlinux	EXPORT_SYMBOL
>   0x00000000	copy_user_generic_unrolled	vmlinux	EXPORT_SYMBOL
>
>   If I ignore these and go on to boot the resulting kernel. I am left with:
>
>   ERROR: Unable to find root device 'UUID=<omitted>'
>
>   I am then dropped to a recovery shell where the keyboard is not answering.
>
>   The UUID is valid and is the same being used when I boot the HEAD of the linux
>   tree. In fact, I am sending this message from the same machine, running
>   that very kernel.
>   Prior to failing to find the root device, the kernel prints a series of
>   messages for each of the above symbol:
>
>   scsi_mod: no symbol version for <foo>
>   scsi_mod: Unknown symbol <foo> (err -22)
>
>   Therefore, I suspect that the missing CRC messages are linked to the issue.
>
>   I have git bisected this behavior to the patch 784d5699edd.
>
>   Simply reverting this patch results in different issues.
>   The errors I am having due to no CRC are fixed but I receive another error at
>   boot time:
>   failed command: WRITE LOG DMA EXT
>
>   I mostly asking for help as to what to do next.
>   Could someone point me toward documentation that would help me understand and
>   fix this issue?
I just wanted to add that I tried to compile the kernel on a different 
machine and I received
the same warnings and error messages.
They both run Archlinux, so this may be a factor.

Quentin

Re: Unable to boot linux-next build

[Alban Browaeys]

Le dimanche 11 septembre 2016 à 20:00 +0200, Quentin Lambert a écrit :
> 
> On 09/09/2016 10:31, Quentin Lambert wrote:
> >   I have been trying to build and boot the last version available
> > on linux-next.
> >   During the build I am being prompted with "has no CRC!" warnings
> > for a bunch
> >   of modules.

There is a thread about this issue, with patches on the list (lkml)
titled "[GIT PULL] kbuild changes for v4.9-rc1" starting on Mon, 17 Oct
2016, with initial author "Michal Marek".

To sum up, you could disable CONFIG_MODVERSIONS in your config or
downgrade binutils to <= 2.26 or revert 784d5699eddc ("x86: move
exports to actual definitions")  or apply the two patches from the
above thread if on x86:

- first one : from Nicholas Piggin for all archs:
http://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1250340.htm
l

- second from Adam Borowski , x86 specific bits :
http://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1250599.htm
l