Some suggestions for improvements in GRUB

Some suggestions for improvements in GRUB

[Karthik Krishna]

[-- Attachment #1: Type: text/plain, Size: 1759 bytes --]

Hi there,

I am an engineering student from India. I have been using Linux Distros for
some time. With its varied uses i have come to like GRUB very much. But i
wish to say that there are certain things that are not user friendly and
needs a bit of change.

First of all, if a drive is deleted or edited using any partition manager,
GRUB does not know anything about it. And so in the ensuing part of
booting, it shows a boot error. i have been seeing this as my friends keep
coming to me to "FIX" their boot. The problem lies in GRUB's steep learning
curve in the  fixing department. when the grub error screen comes up, one
has to come up with a plan to go to the internet and see the required grub
error codes and relevant data to fix it. Thereby restoring grubs working.
But the thing is you need an another PC/laptop with a working internet to
see the codes or relevant data. i have been doing this for some time. i see
that people who wish to use Linux have a hard time using it fully,
especially when it comes to its boot-loader.

The onset of boot-loader problems really bogs down the motivation to move
forward.

So I wish to develop GRUB a bit so that it may be easy to use even in the
case of error handling. I just dont know where to start. I know assembly, a
bit of c and c++. I have heard of nasm and used gcc. So how do i develop
it? I wanted to make a code that automatically detects partitions and
checks it with a library of partitions to see if it has been changed or
not. So that way the user is given the freedom of not doing anything rather
letting GRUB doing the work for the user. i can develop the code using a
simple algorithm. i want to make grub more smarter when it comes to
handling errors and freeing the user.

Karthik Krishna

[-- Attachment #2: Type: text/html, Size: 1838 bytes --]

Re: Some suggestions for improvements in GRUB

[Chris Murphy]

On Apr 22, 2012, at 12:49 PM, Karthik Krishna wrote:

> Hi there,
> 
> I am an engineering student from India. I have been using Linux Distros for some time. With its varied uses i have come to like GRUB very much. But i wish to say that there are certain things that are not user friendly and needs a bit of change.
> 
> First of all, if a drive is deleted or edited using any partition manager, GRUB does not know anything about it. And so in the ensuing part of booting, it shows a boot error. i have been seeing this as my friends keep coming to me to "FIX" their boot. The problem lies in GRUB's steep learning curve in the  fixing department. when the grub error screen comes up, one has to come up with a plan to go to the internet and see the required grub error codes and relevant data to fix it. Thereby restoring grubs working.  But the thing is you need an another PC/laptop with a working internet to see the codes or relevant data. i have been doing this for some time. i see that people who wish to use Linux have a hard time using it fully, especially when it comes to its boot-loader.
> 
> The onset of boot-loader problems really bogs down the motivation to move forward.
> 
> So I wish to develop GRUB a bit so that it may be easy to use even in the case of error handling. I just dont know where to start. I know assembly, a bit of c and c++. I have heard of nasm and used gcc. So how do i develop it? I wanted to make a code that automatically detects partitions and checks it with a library of partitions to see if it has been changed or not. So that way the user is given the freedom of not doing anything rather letting GRUB doing the work for the user. i can develop the code using a simple algorithm. i want to make grub more smarter when it comes to handling errors and freeing the user.


Brilliant. 

I've been wondering if it's possible to get the existing grub-mkconfig script and os-prober code, etc. wrapped into a GRUB module, and have it dynamically create a grub.cfg at boot time. It would not need to write out a grub.cfg, but use it as a "master" cfg, while incorporating any located grub.cfg on-disk in a separate menu. That way, there's both. 

But I agree, I know way way way way too much about GRUB to get reliable booting. mkconfig is leap years beyond the asylum of having to hand write menu.lst but even more automatic is better.

For UEFI systems, I think GRUB2 is way too complicated, IMO. "All" that's needed there is a way to make up for the fact so many UEFI implementations have no interface to choose the bootloader for other OS's. And a prettier menu, kinda like rEFIt (now rEFInd) or the way Apple does it, for basic mode UI would be so much of a better experience. Fine, have an advanced menu to get to for people who need a GRUB prompt, but I don't think most users need to be interacting with GRUB as much as they have been historically.


Chris Murphy

Re: Some suggestions for improvements in GRUB

[Vladimir 'φ-coder/phcoder' Serbinenko]

[-- Attachment #1: Type: text/plain, Size: 3616 bytes --]

On 23.04.2012 07:23, Chris Murphy wrote:
> On Apr 22, 2012, at 12:49 PM, Karthik Krishna wrote:
>
>> Hi there,
>>
>> I am an engineering student from India. I have been using Linux Distros for some time. With its varied uses i have come to like GRUB very much. But i wish to say that there are certain things that are not user friendly and needs a bit of change.
>>
>> First of all, if a drive is deleted or edited using any partition manager, GRUB does not know anything about it. And so in the ensuing part of booting, it shows a boot error. i have been seeing this as my friends keep coming to me to "FIX" their boot. The problem lies in GRUB's steep learning curve in the  fixing department. when the grub error screen comes up, one has to come up with a plan to go to the internet and see the required grub error codes and relevant data to fix it. Thereby restoring grubs working.  But the thing is you need an another PC/laptop with a working internet to see the codes or relevant data. i have been doing this for some time. i see that people who wish to use Linux have a hard time using it fully, especially when it comes to its boot-loader.
>>
>> The onset of boot-loader problems really bogs down the motivation to move forward.
>>
>> So I wish to develop GRUB a bit so that it may be easy to use even in the case of error handling. I just dont know where to start. I know assembly, a bit of c and c++. I have heard of nasm and used gcc. So how do i develop it? I wanted to make a code that automatically detects partitions and checks it with a library of partitions to see if it has been changed or not. So that way the user is given the freedom of not doing anything rather letting GRUB doing the work for the user. i can develop the code using a simple algorithm. i want to make grub more smarter when it comes to handling errors and freeing the user.
>
> Brilliant. 
>
> I've been wondering if it's possible to get the existing grub-mkconfig script and os-prober code, etc. wrapped into a GRUB module, and have it dynamically create a grub.cfg at boot time. It would not need to write out a grub.cfg, but use it as a "master" cfg, while incorporating any located grub.cfg on-disk in a separate menu. That way, there's both. 
See os-prober.cfg
> But I agree, I know way way way way too much about GRUB to get reliable booting. mkconfig is leap years beyond the asylum of having to hand write menu.lst but even more automatic is better.
>
> For UEFI systems, I think GRUB2 is way too complicated, IMO. "All" that's needed there is a way to make up for the fact so many UEFI implementations have no interface to choose the bootloader for other OS's. And a prettier menu, kinda like rEFIt (now rEFInd) or the way Apple does it, for basic mode UI would be so much of a better experience. Fine, have an advanced menu to get to for people who need a GRUB prompt, but I don't think most users need to be interacting with GRUB as much as they have been historically.
The task of loading kernel into memory and launching it (avoiding
firmware bugs all the time) is almost the same on EFI and BIOS, you
can't just remove it and leave only menu.
The reason while such OS choosers are so "simple" is that they don't
handle real OS loading at all, that's all they are: OS-choosers. They
don't load kernels. An equivalent on BIOS would be mbldr, not GRUB
>
> Chris Murphy
> _______________________________________________
> Grub-devel mailing list
> Grub-devel@gnu.org
> https://lists.gnu.org/mailman/listinfo/grub-devel
>


-- 
Regards
Vladimir 'φ-coder/phcoder' Serbinenko



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 294 bytes --]

Re: Some suggestions for improvements in GRUB

[Vladimir 'φ-coder/phcoder' Serbinenko]

[-- Attachment #1: Type: text/plain, Size: 2624 bytes --]

On 22.04.2012 20:49, Karthik Krishna wrote:
> Hi there,
>
> I am an engineering student from India. I have been using Linux
> Distros for some time. With its varied uses i have come to like GRUB
> very much. But i wish to say that there are certain things that are
> not user friendly and needs a bit of change.
>
Such presentations are redundant. We treat the patches all the same
independently of who you are.
> First of all, if a drive is deleted or edited using any partition
> manager, GRUB does not know anything about it. And so in the ensuing
> part of booting, it shows a boot error. i have been seeing this as my
> friends keep coming to me to "FIX" their boot. The problem lies in
> GRUB's steep learning curve in the  fixing department. when the grub
> error screen comes up, one has to come up with a plan to go to the
> internet and see the required grub error codes and relevant data to
> fix it. Thereby restoring grubs working.  But the thing is you need an
> another PC/laptop with a working internet to see the codes or relevant
> data. i have been doing this for some time. i see that people who wish
> to use Linux have a hard time using it fully, especially when it comes
> to its boot-loader.
GRUB2 has no error codes at all.
>
> The onset of boot-loader problems really bogs down the motivation to
> move forward.
>
> So I wish to develop GRUB a bit so that it may be easy to use even in
> the case of error handling. I just dont know where to start. I know
> assembly, a bit of c and c++. I have heard of nasm and used gcc. So
> how do i develop it? I wanted to make a code that automatically
> detects partitions and checks it with a library of partitions to see
> if it has been changed or not. So that way the user is given the
> freedom of not doing anything rather letting GRUB doing the work for
> the user. i can develop the code using a simple algorithm. i want to
> make grub more smarter when it comes to handling errors and freeing
> the user.
>
In most places GRUB2 uses UUIDs which shouldn't change on partition
modifications. The only place another form is used is on prefix
specification and platform-specific install (embedding zone, EFI
partition, PreP and so on).
It's possible to add --no-use-partition-number to grub-install for the
first part, the patch is 3-5 lines, but it increases kernel size.
As for the second, there is no way around it as on most platforms it's
tied in to platform limitations.
You need to rerun grub-install after important partitioning changes.

-- 
Regards
Vladimir 'φ-coder/phcoder' Serbinenko



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 294 bytes --]