Skip to contentSkip to conversation

Partially defer some of the async stuff to the next release

DISCUSSION Arjan van de Ven 4 participants 7 messages Jan 9, 2009

lore
Share & tools
lore
#0 lore raw

Hi Linus,

I think it's better that we defer some of the asynchronous boot stuff to 2.6.30 so that it can bake in -next for a bit. I would like to keep the core infrastructure in place for 29, so that the various subsystem patches etc can just use it in -next without generating dependencies in the patch flow.

for both libata and the inode delete very simple things can be done to fix them, but I would feel a lot more comfortable giving these a ride in -next.

The following changes since commit 2150edc6c5cf00f7adb54538b9ea2a3e9cedca3f: Linus Torvalds (1):
Merge branch 'for_linus' of git://git.kernel.org/.../tytso/ext4

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/arjan/linux-2.6-async-undo.git
master

Arjan van de Ven (3):
Revert "fastboot: Make libata initialization even more async" Revert "fastboot: make the libata port scan asynchronous" partial revert of asynchronous inode delete

 drivers/ata/libata-core.c |   96

+++++++++++++++++++++------------------------
fs/inode.c | 19 +++------ 2 files changed, 51
insertions(+), 64 deletions(-)

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#1 lore raw
Quoted 10 lines · Arjan van de Ven

I think it's better that we defer some of the asynchronous boot stuff to 2.6.30 so that it can bake in -next for a bit. I would like to keep the core infrastructure in place for 29, so that the various subsystem patches etc can just use it in -next without generating dependencies in the patch flow.

for both libata and the inode delete very simple things can be done to fix them, but I would feel a lot more comfortable giving these a ride in -next.

Can we rather make this something that people can enable with a "fastboot" option and is disabled by default? At least that way the ATA people can see the problem, and don't just forget about it, since we want it fixed.

Linus

#2 lore raw
Quoted 10 lines · On Fri, 9 Jan 2009 12:42:34 -0800 (PST) Linus Torvalds…
Quoted 7 lines · Arjan van de Ven

I think it's better that we defer some of the asynchronous boot stuff to 2.6.30 so that it can bake in -next for a bit. I would like to keep the core infrastructure in place for 29, so that the various subsystem patches etc can just use it in -next without generating dependencies in the patch flow.

Quoted 8 lines
Quoted 3 lines

for both libata and the inode delete very simple things can be done to fix them, but I would feel a lot more comfortable giving these a ride in -next.

Can we rather make this something that people can enable with a "fastboot" option and is disabled by default? At least that way the ATA people can see the problem, and don't just forget about it, since we want it fixed.

ok sure, that's not hard. I'll get that going, hopefully before the end of the day

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#3 lore raw
In reply to Linus Torvalds
Quoted 3 lines

Can we rather make this something that people can enable with a "fastboot" option and is disabled by default? At least that way the ATA people can see the problem, and don't just forget about it, since we want it fixed.

Makes some sense - although the inode one should just take a hike at this point. Can the fastboot option live under staging perhaps - its analogous to staging drivers although clearly the code doesn't go there in this case ?

The trouble otherwise is "fast boot" sounds like something good for people to turn on, but right now it's the reverse.

Alan

#4 lore raw
Quoted 5 lines · On Fri, 9 Jan 2009 21:22:01 +0000 Alan Cox <[email protected]> wrote:
Quoted 4 lines

Can we rather make this something that people can enable with a "fastboot" option and is disabled by default? At least that way the ATA people can see the problem, and don't just forget about it, since we want it fixed.

Quoted 7 lines

Makes some sense - although the inode one should just take a hike at this point. Can the fastboot option live under staging perhaps - its analogous to staging drivers although clearly the code doesn't go there in this case ?

The trouble otherwise is "fast boot" sounds like something good for people to turn on, but right now it's the reverse.

I just coded it as a kernel command line option only

that's very much a "I know what I'm doing" kind of thng.

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#5 lore raw
In reply to Alan Cox
Quoted 11 lines · Alan Cox
Quoted 3 lines

Can we rather make this something that people can enable with a "fastboot" option and is disabled by default? At least that way the ATA people can see the problem, and don't just forget about it, since we want it fixed.

Makes some sense - although the inode one should just take a hike at this point. Can the fastboot option live under staging perhaps - its analogous to staging drivers although clearly the code doesn't go there in this case ?

The trouble otherwise is "fast boot" sounds like something good for people to turn on, but right now it's the reverse.

isn't this like OPTIMIZE_INLINE?

David Lang

#6 lore raw
In reply to Linus Torvalds
Quoted 10 lines · On Fri, 9 Jan 2009 12:42:34 -0800 (PST) Linus Torvalds…
Quoted 7 lines · Arjan van de Ven

I think it's better that we defer some of the asynchronous boot stuff to 2.6.30 so that it can bake in -next for a bit. I would like to keep the core infrastructure in place for 29, so that the various subsystem patches etc can just use it in -next without generating dependencies in the patch flow.

Quoted 8 lines
Quoted 3 lines

for both libata and the inode delete very simple things can be done to fix them, but I would feel a lot more comfortable giving these a ride in -next.

Can we rather make this something that people can enable with a "fastboot" option and is disabled by default? At least that way the ATA people can see the problem, and don't just forget about it, since we want it fixed.

if you pull this you get the patch that removes the inode part for now, and adds a "fastboot" kernel commandline option to enable the async behavior.

The following changes since commit 7c51d57e9d7fbce89f79c41dc8da383101dbe9c6: Linus Torvalds (1):

        Merge git://git.infradead.org/mtd-2.6

are available in the git repository at:

ssh://master.kernel.org/pub/scm/linux/kernel/git/arjan/linux-2.6-async-2 master

Arjan van de Ven (2):
partial revert of asynchronous inode delete
async: make async a command line option for now

 fs/inode.c     |   19 +++++++------------
 kernel/async.c |   16 ++++++++++++++--
 2 files changed, 21 insertions(+), 14 deletions(-)
Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org