Skip to contentSkip to conversation

71715 -> Not applicable.

U-Boot Albert ARIBAUD 3 participants 11 messages Nov 18, 2010 – Nov 20, 2010

lore
Share & tools
lore
#0 lore raw

Yay, my first todo in patchwork! :)

<http://patchwork.ozlabs.org/patch/71715/>

It was discussed there:

<http://www.mail-archive.com/[email protected]/msg28003.html>

AIUI, the discussion was about asm-generic/unaligned.h rather than about the patch itself. Personally I think that the patch could have been applied, and *then* asm-generic/unaligned.h could have been brought in sync with its Linux counterpart.

Anyway this patch requires a rebase now as the whole arch organization has moved around. Ergo: not applicable; if the patch is needed, I'd like it to be resubmitted based on current master.

Note: I am posting this to the list because we're only starting using patchwork, so maybe we should share our experience for the moment; I don't know yet what patchwork action is worthy of a post on the list and what is not. For all I know, maybe patchwork itself is going to post something right after I set the "not applicable" status...

In any case, *do not* blindly take this message as the Right Way To Do Things With Patchwork! I don't want to start a wrong habit.

Amicalement,

Signature · 1 line
Albert.
#1 lore raw

Dear Albert ARIBAUD,

Quoted 6 lines · In message <[email protected]> you wrote:

Note: I am posting this to the list because we're only starting using patchwork, so maybe we should share our experience for the moment; I don't know yet what patchwork action is worthy of a post on the list and what is not. For all I know, maybe patchwork itself is going to post something right after I set the "not applicable" status...

No, I don't think so. Patchwork does not send any e-mails by itself. When you change the state to "Changes Required" or "Not Applicable" or such you are supposed to post a followup on the mailinglist and inform the submitter of your action. Patchwork does not replace the mailing list, it just helps keeping an overview over the patches, and who is doing what.

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Overdrawn?  But I still have checks left!
#2 lore raw
In reply to Albert ARIBAUD
Quoted 8 lines · On Thursday, November 18, 2010 16:31:58 Albert ARIBAUD wrote:

Yay, my first todo in patchwork! :)

<http://patchwork.ozlabs.org/patch/71715/>

It was discussed there:

<http://www.mail-archive.com/[email protected]/msg28003.html>

Quoted 8 lines

AIUI, the discussion was about asm-generic/unaligned.h rather than about the patch itself. Personally I think that the patch could have been applied, and *then* asm-generic/unaligned.h could have been brought in sync with its Linux counterpart.

Anyway this patch requires a rebase now as the whole arch organization has moved around. Ergo: not applicable; if the patch is needed, I'd like it to be resubmitted based on current master.

it's trivial to rebase the patch yourself. you're the arm maintainer, and this patch is for the core arm.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20101118/da62b7c0/attachment.pgp

#3 lore raw

Le 18/11/2010 23:00, Mike Frysinger a ?crit :

Quoted 7 lines
Quoted 3 lines

Anyway this patch requires a rebase now as the whole arch organization has moved around. Ergo: not applicable; if the patch is needed, I'd like it to be resubmitted based on current master.

it's trivial to rebase the patch yourself. you're the arm maintainer, and this patch is for the core arm.
-mike

Yes, it is trivial, and yes, I am the arm maintainer (well, technically not until the next release is out), and as the arm maintainer I see patch which has been left dormant since january without adverse effects and without a ping (at least, none that I could see), which makes me wonder how strongly it is really wished for; a resubmit will be a sign that it is.

Besides, about half the archs do not use asm-generic/unaligned, and half do; a resubmit on the list will be the occasion, for example, to discuss whether all arches should use it.

Amicalement,

Signature · 1 line
Albert.
#4 lore raw
Quoted 2 lines · On Thursday, November 18, 2010 17:21:49 Albert ARIBAUD wrote:

Besides, about half the archs do not use asm-generic/unaligned, and half do

that's because half the arches implemented unaligned.h whilst only thinking of themselves. the other half benefited from my work of thinking of everyone.

Quoted 1 line

a resubmit on the list will be the occasion

i dont use/care about arm. i wrote the patch because i wanted to be nice. if you cant be bothered to clean up arm cruft, then feel free to mark the patch as "pending due to maintainer laziness".

Quoted 1 line

for example, to discuss whether all arches should use it.

all arches should use it. whether the maintainers can be bothered to fix their headers is a different question.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20101118/d196d3f7/attachment.pgp

#5 lore raw

Le 19/11/2010 00:13, Mike Frysinger a ?crit :

Quoted 13 lines · On Thursday, November 18, 2010 17:21:49 Albert ARIBAUD wrote:
Quoted 2 lines

Besides, about half the archs do not use asm-generic/unaligned, and half do

that's because half the arches implemented unaligned.h whilst only thinking of themselves. the other half benefited from my work of thinking of everyone.

Quoted 1 line

a resubmit on the list will be the occasion

i dont use/care about arm. i wrote the patch because i wanted to be nice. if you cant be bothered to clean up arm cruft, then feel free to mark the patch as "pending due to maintainer laziness".

Quoted 5 lines
Quoted 1 line

for example, to discuss whether all arches should use it.

all arches should use it. whether the maintainers can be bothered to fix their headers is a different question.
-mike

My question was about the technical merits of the patch: why should all arches use it? What does it improve at a performance, maintenability, or other technical level ?

Amicalement,

Signature · 1 line
Albert.
#6 lore raw
Quoted 15 lines · On Friday, November 19, 2010 00:59:47 Albert ARIBAUD wrote:

Le 19/11/2010 00:13, Mike Frysinger a ?crit :

Quoted 14 lines · On Thursday, November 18, 2010 17:21:49 Albert ARIBAUD wrote:
Quoted 2 lines

Besides, about half the archs do not use asm-generic/unaligned, and half do

that's because half the arches implemented unaligned.h whilst only thinking of themselves. the other half benefited from my work of thinking of everyone.

Quoted 1 line

a resubmit on the list will be the occasion

i dont use/care about arm. i wrote the patch because i wanted to be nice. if you cant be bothered to clean up arm cruft, then feel free to mark the patch as "pending due to maintainer laziness".

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

for example, to discuss whether all arches should use it.

all arches should use it. whether the maintainers can be bothered to fix their headers is a different question.

My question was about the technical merits of the patch: why should all arches use it? What does it improve at a performance, maintenability, or other technical level ?

obviously maintenance is improved since only one header needs to be maintained and it isnt an arm one
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20101119/9c6b683a/attachment.pgp

#7 lore raw

Le 19/11/2010 08:13, Mike Frysinger a ?crit :

Quoted 21 lines · On Friday, November 19, 2010 00:59:47 Albert ARIBAUD wrote:
Quoted 20 lines

Le 19/11/2010 00:13, Mike Frysinger a ?crit :

Quoted 18 lines · On Thursday, November 18, 2010 17:21:49 Albert ARIBAUD wrote:
Quoted 2 lines

Besides, about half the archs do not use asm-generic/unaligned, and half do

that's because half the arches implemented unaligned.h whilst only thinking of themselves. the other half benefited from my work of thinking of everyone.

Quoted 1 line

a resubmit on the list will be the occasion

i dont use/care about arm. i wrote the patch because i wanted to be nice. if you cant be bothered to clean up arm cruft, then feel free to mark the patch as "pending due to maintainer laziness".

Quoted 1 line

for example, to discuss whether all arches should use it.

all arches should use it. whether the maintainers can be bothered to fix their headers is a different question.

Quoted 7 lines
Quoted 3 lines

My question was about the technical merits of the patch: why should all arches use it? What does it improve at a performance, maintenability, or other technical level ?

obviously maintenance is improved since only one header needs to be maintained and it isnt an arm one
-mike

Maintenance being about changes, I did a git log on both include/asm-generic/unaligned.h and arch/arm/include/asm/unaligned.h. Each has exactly one commit: the generic one when creating the file, the arm one when the directories were rearranged. Very little changes either way.

I do understand the benefit in overall maintenability of having a common situation for all archs. But I don't think there is an increase of maintenability per se in, basically, adding a level of #include.

If the goal is 'have all arches use a single unaligned.h' -- then fine, let us *remove* the unaligned.h files in arches and refer to the generic one instead. That, indeed, would increase maintenability by *reducing* the number of files while *not* adding any complexity. I'll happily ack such a change.

In any case, this is not specifically an 'arm' topic: it touches all arches albeit lightly -- as you point out, Mike, this is a generic change. So I'd rather see a patchset to fix this globally and for good for every arch.

Amicalement,

Signature · 1 line
Albert.
#8 lore raw
Quoted 5 lines · On Friday, November 19, 2010 05:50:20 Albert ARIBAUD wrote:

If the goal is 'have all arches use a single unaligned.h' -- then fine, let us *remove* the unaligned.h files in arches and refer to the generic one instead. That, indeed, would increase maintenability by *reducing* the number of files while *not* adding any complexity. I'll happily ack such a change.

that really isnt going to happen. files expect asm/foo.h. just like Linux.

Quoted 4 lines

In any case, this is not specifically an 'arm' topic: it touches all arches albeit lightly -- as you point out, Mike, this is a generic change. So I'd rather see a patchset to fix this globally and for good for every arch.

i have 0 incentive or desire to fix other arches. so that isnt going to happen by me either.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20101119/213fbf81/attachment.pgp

#9 lore raw

Le 19/11/2010 11:51, Mike Frysinger a ?crit :

Quoted 8 lines · On Friday, November 19, 2010 05:50:20 Albert ARIBAUD wrote:
Quoted 5 lines

If the goal is 'have all arches use a single unaligned.h' -- then fine, let us *remove* the unaligned.h files in arches and refer to the generic one instead. That, indeed, would increase maintenability by *reducing* the number of files while *not* adding any complexity. I'll happily ack such a change.

that really isnt going to happen. files expect asm/foo.h. just like Linux.

Then why not symlinks?

Quoted 8 lines
Quoted 4 lines

In any case, this is not specifically an 'arm' topic: it touches all arches albeit lightly -- as you point out, Mike, this is a generic change. So I'd rather see a patchset to fix this globally and for good for every arch.

i have 0 incentive or desire to fix other arches. so that isnt going to happen by me either.
-mike

If no one -- including the submitter -- expresses an incentive for this change, it most probably means it is not needed. I shall thus drop this patch; if anyone else feels it is needed, they can simply submit an up-to-date patch on the list.

Amicalement,

Signature · 1 line
Albert.
#10 lore raw
Quoted 8 lines · On Friday, November 19, 2010 06:56:38 Albert ARIBAUD wrote:

Le 19/11/2010 11:51, Mike Frysinger a ?crit :

Quoted 7 lines · On Friday, November 19, 2010 05:50:20 Albert ARIBAUD wrote:
Quoted 5 lines

If the goal is 'have all arches use a single unaligned.h' -- then fine, let us *remove* the unaligned.h files in arches and refer to the generic one instead. That, indeed, would increase maintenability by *reducing* the number of files while *not* adding any complexity. I'll happily ack such a change.

Quoted 4 lines
Quoted 2 lines

that really isnt going to happen. files expect asm/foo.h. just like Linux.

Then why not symlinks?

both build systems have actively moved away from symlinks. they turn out to be way too problematic in practice.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20101120/12cc0b81/attachment.pgp