Skip to contentSkip to conversation

MSI rework

PATCH Eric W. Biederman 4 participants 9 messages Mar 22, 2007 – Mar 26, 2007 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): 1174560686.307511.956605711793.qpush@cradle

#1 lore raw
Quoted 6 lines · Michael Ellerman

This is my series to rework the generic MSI code into something we can use on powerpc[1].

I've tried as much as possible not to change the semantics for other archs, but there's a few little changes. I think they're all OK in their own right.

Quoted 7 lines

I don't have any serious hardware to test on, but my little x86_64 box with an e1000 using MSI still works with these changes. I've also got MSI working on a powerpc blade with a tg3. I haven't tested MSI-X _at all_.

I've also tested on the blade with a debug hack to make the MSI case allocate/free 8 MSIs, but only use the last one, just to exercise the n > 1 case a little. All seems to work fine.

Generally I think this looks good. However there is a lot here, and some of it potentially at least has some pretty subtle side effects.

So reviewing all of these patches at once is almost certain to cause something important to be missed :(

Can we slow this process down a few days by taking this one logical chunk at a time?

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

Eric

#2 lore raw
Quoted 22 lines · Eric W. Biederman
Quoted 6 lines · Michael Ellerman

This is my series to rework the generic MSI code into something we can use on powerpc[1].

I've tried as much as possible not to change the semantics for other archs, but there's a few little changes. I think they're all OK in their own right.

Quoted 7 lines

I don't have any serious hardware to test on, but my little x86_64 box with an e1000 using MSI still works with these changes. I've also got MSI working on a powerpc blade with a tg3. I haven't tested MSI-X _at all_.

I've also tested on the blade with a debug hack to make the MSI case allocate/free 8 MSIs, but only use the last one, just to exercise the n > 1 case a little. All seems to work fine.

Generally I think this looks good. However there is a lot here, and some of it potentially at least has some pretty subtle side effects.

So reviewing all of these patches at once is almost certain to cause something important to be missed :(

Quoted 7 lines

Can we slow this process down a few days by taking this one logical chunk at a time?

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

I agree, care to break these down into a smaller series of patches that can go into -mm for testing?

thanks,

greg k-h

#3 lore raw
Quoted 7 lines
Quoted 4 lines

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

I agree, care to break these down into a smaller series of patches that can go into -mm for testing?

I don't see the point in breaking the serie... you can bisect half way through if necessary... it's made of small patches that are done, afaik, in such a way that the whole thing should still work at any level in the serie.

The serie just expresses the dependency between them.

Ben.

#4 lore raw
Quoted 8 lines · Benjamin Herrenschmidt
Quoted 7 lines
Quoted 4 lines

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

I agree, care to break these down into a smaller series of patches that can go into -mm for testing?

Quoted 6 lines

I don't see the point in breaking the serie... you can bisect half way through if necessary... it's made of small patches that are done, afaik, in such a way that the whole thing should still work at any level in the serie.

The serie just expresses the dependency between them.

Ok, then which patches in the series should be acceptable to take right now for 2.6.22? The "clean up the BUG" ones?

thanks,

greg k-h

#5 lore raw
Quoted 3 lines

Ok, then which patches in the series should be acceptable to take right now for 2.6.22? The "clean up the BUG" ones?

I'd say 1...8 (BUG cleanups and removing msi_cache) looks to be on the safe side to me, though of course, some review by Eric & a bit of testing would be welcome regardless :-)

The others could go into -mm right away I suppose (or rather as soon as Eric gets a chance to verify they don't break his test setups as I think he can test MSI-X) and we can pour them into mainline one by one if necessary as they get hammered / verified.

Cheers,
Ben.

#6 lore raw
In reply to Greg KH
Quoted 14 lines · Greg KH
Quoted 13 lines · Benjamin Herrenschmidt
Quoted 7 lines
Quoted 4 lines

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

I agree, care to break these down into a smaller series of patches that can go into -mm for testing?

I don't see the point in breaking the serie... you can bisect half way through if necessary... it's made of small patches that are done, afaik, in such a way that the whole thing should still work at any level in the serie.

Quoted 4 lines
Quoted 1 line

The serie just expresses the dependency between them.

Ok, then which patches in the series should be acceptable to take right now for 2.6.22? The "clean up the BUG" ones?

The series is already very verbose, I don't think I can split most of them any smaller without producing an unbuildable kernel.

I think 1 up to and including 11 are safe as houses, they shouldn't have any effect other than to clean up the code.

The rest make functional changes, but they're all quite small, self contained, and easily bisectable. I'd certainly like Eric to have a look at them, but at some point I think we're just going to have to bite the bullet and merge them, and see what we get in the way of bug reports.

cheers

Signature · 8 lines
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person
#7 lore raw
Quoted 15 lines · Michael Ellerman
Quoted 14 lines · Greg KH
Quoted 13 lines · Benjamin Herrenschmidt
Quoted 7 lines
Quoted 4 lines

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

I agree, care to break these down into a smaller series of patches that can go into -mm for testing?

I don't see the point in breaking the serie... you can bisect half way through if necessary... it's made of small patches that are done, afaik, in such a way that the whole thing should still work at any level in the serie.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

The serie just expresses the dependency between them.

Ok, then which patches in the series should be acceptable to take right now for 2.6.22? The "clean up the BUG" ones?

The series is already very verbose, I don't think I can split most of them any smaller without producing an unbuildable kernel.

That wasn't the request.

Quoted 7 lines

I think 1 up to and including 11 are safe as houses, they shouldn't have any effect other than to clean up the code.

The rest make functional changes, but they're all quite small, self contained, and easily bisectable. I'd certainly like Eric to have a look at them, but at some point I think we're just going to have to bite the bullet and merge them, and see what we get in the way of bug reports.

What I wanted was the patches organized into functional groups that were small enough to review as a unit. (Feed the existing patches slower please).

This seems to be to much change to read and review as a unit, I just get bleary eyed, and start to get confused.

So far I have found one subtle bug. Where admittedly my code wasn't as obvious as it could be and you were proposing to use an irq that had already been freed.

What I had hoped we can do is you would send a handful at a time I would review them. Then we could get the next handful. I expect doing it that way it should take about a week to get through them all.

I guess I can try going through the review that way as well. Pick a subset of what you have sent and review it very carefully, and the next day pick a different subset.

I have been sufficiently active in this code lately that I think I can do a good review, and I want to. Unfortunately I'm only human and a good review is nearly as much work as writing the patches myself which means it takes time.

Eric

#8 lore raw
Quoted 46 lines · Eric W. Biederman
Quoted 15 lines · Michael Ellerman
Quoted 14 lines · Greg KH
Quoted 13 lines · Benjamin Herrenschmidt
Quoted 7 lines
Quoted 4 lines

i.e. First the simple bug fixes that should purely be restructure of msi.c with no affect on anything outside of it.

And then get into the architecture enhancements.

I agree, care to break these down into a smaller series of patches that can go into -mm for testing?

I don't see the point in breaking the serie... you can bisect half way through if necessary... it's made of small patches that are done, afaik, in such a way that the whole thing should still work at any level in the serie.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

The serie just expresses the dependency between them.

Ok, then which patches in the series should be acceptable to take right now for 2.6.22? The "clean up the BUG" ones?

The series is already very verbose, I don't think I can split most of them any smaller without producing an unbuildable kernel.

That wasn't the request.

Quoted 7 lines

I think 1 up to and including 11 are safe as houses, they shouldn't have any effect other than to clean up the code.

The rest make functional changes, but they're all quite small, self contained, and easily bisectable. I'd certainly like Eric to have a look at them, but at some point I think we're just going to have to bite the bullet and merge them, and see what we get in the way of bug reports.

What I wanted was the patches organized into functional groups that were small enough to review as a unit. (Feed the existing patches slower please).

This seems to be to much change to read and review as a unit, I just get bleary eyed, and start to get confused.

So far I have found one subtle bug. Where admittedly my code wasn't as obvious as it could be and you were proposing to use an irq that had already been freed.

Quoted 7 lines

What I had hoped we can do is you would send a handful at a time I would review them. Then we could get the next handful. I expect doing it that way it should take about a week to get through them all.

I guess I can try going through the review that way as well. Pick a subset of what you have sent and review it very carefully, and the next day pick a different subset.

OK. For starters, do you want to review the first eleven as I've sent them already, that saves spamming everyone again.

If you're OK with those eleven, then I'll send the remaining 10 or so later in the week, broken up into (sort-of) functional groups.

cheers

Signature · 8 lines
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person
#9 lore raw
Quoted 5 lines · Michael Ellerman

OK. For starters, do you want to review the first eleven as I've sent them already, that saves spamming everyone again.

If you're OK with those eleven, then I'll send the remaining 10 or so later in the week, broken up into (sort-of) functional groups.

Sounds like a decent plan. At a quick glance there is a reasonable chance I'm going to run into issues with patch 9 or 10. Let me get as far as I can with the acks and then we can see where we go with the next round of patches.

Eric