On Sun, Nov 3, 2013 at 7:16 PM, Vladimir 'φ-coder/phcoder' Serbinenko
<phcoder@gmail.com> wrote:
> On 04.11.2013 02:10, Javier Vasquez wrote:
>> On 11/1/13, Vladimir 'φ-coder/phcoder' Serbinenko <phcoder@gmail.com> wrote:
>>> On 01.11.2013 22:26, Aleš Nesrsta wrote:
>>>>
>>>>
>>>> Dne 29.10.2013 19:46, Vladimir 'φ-coder/phcoder' Serbinenko napsal(a):
>>>>> On 29.10.2013 19:35, Aleš Nesrsta wrote:
>>>>>> Only short note below...
>>>>>>
>>>>>> Dne 27.10.2013 23:43, Vladimir 'φ-coder/phcoder' Serbinenko napsal(a):
>>>>>>> On 27.10.2013 21:03, Aleš Nesrsta wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>> 2.
>>>>>>>> I don't see loading of OHCI module in debug output !
>>>>>>>> Do you really have this module included in your image?
>>>>>>> I have to second this: OHCI module if loaded and there is at least one
>>>>>>> PCI device available it will show "pciid = %x" message.
>>>>>>> Can you modify grub_real_dprintf in misc.c to:
>>>>>>> if (!debug)
>>>>>>>     debug = "all";
>>>>>>>
>>>>>>> So we can be sure that the message wasn't simply lost due to late
>>>>>>> setting of debug variable.
>>>>>>
>>>>>>
>>>>>> AFAIK, dl.c module should display for each loaded module something like
>>>>>> this (when debug=all as in Javier case):
>>>>>> "kern/dl.c:630: module name: ehci"
>>>>>> This is missing for OHCI module in debug output sent by Javier, even
>>>>>> grub.cfg command to load OHCI is present on the beginning of debug
>>>>>> output:
>>>>>> ...
>>>>>> script/lexer.c:321: token 288 text [insmod]
>>>>>>
>>>>>> script/script.c:50: malloc 0x8ff1d920
>>>>>>
>>>>>> script/script.c:50: malloc 0x8ff1d900
>>>>>>
>>>>>> script/script.c:163: arglist
>>>>>>
>>>>>> script/script.c:50: malloc 0x8ff1d8d0
>>>>>>
>>>>>> script/lexer.c:321: token 288 text [ohci]
>>>>>> ...
>>>>>>
>>>>>> I.e., from my point of view, it is sure that OHCI module was not loaded
>>>>>> - but there is no debug output which can explain why...
>>>>> It's possible that a module named ohci already loaded.
>>>> Hm. Originally I didn't believe it but you are right. Additionally, it
>>>> is probably the case of the second Javier dump - there is missing dl.c
>>>> message also for loading of EHCI module even it is working - so EHCI had
>>>> to be loaded before insmod command.
>>>>
>>>> Is it there some way how Javier can avoid it?
>>>> I.e., how to load USB drivers at the time when their debug output can be
>>>> captured - ? We need to see debug output of initialization of OHCI
>>>> controller and connected devices...
>>>>
>>> --disk-module=pata to grub-install.
>>>> BR, Ales
>>>>
>>>> _______________________________________________
>>>> Grub-devel mailing list
>>>> Grub-devel@gnu.org
>>>> https://lists.gnu.org/mailman/listinfo/grub-devel
>>>
>>>
>>>
>>
>> Hmm,
>>
>> The following just fails:
>>
>> % sudo grub-install --boot-directory=/boot --target=mipsel-loongson
>> --modules="ehci ohci usb_keyboard" --disk-module=pata --recheck
>> /dev/sda
>> /usr/bin/grub-mkimage: error: cannot open
>> `grub-core/kern/disk.c:196:.mod': No such file or directory.
>>
>> I just compiled/installed grub from git checksum
>> 323de05ba27594861c75eec0b39cb01049a023c9.
>>
>> I suppose I better wait, so that I can provide the right output dump.
>>
> When you do grub_dprintf change you have to do it like:
>   if (!debug)
> #ifdef GRUB_UTIL
>      return;
> #else
>      debug="all";
> #endif


I'll be sending the grub.cfg from the mini-pc later, but now I'm able
to see very little output through the serial interface.  Attached goes
this output.


-- 
Javier.