EH progress summary

DISCUSSION Tejun Heo 2 participants 4 messages Feb 20, 2006 libata

Share & tools

Tejun Heo

author

2006-02-20 03:40 UTC #0 lore raw

Hello, Jeff.

I think we're half-way through the EH changes. New reset mechanism and most of config reorganization are in tree. Currently, the patches which are waiting for your review are...

[1] libata: rename ata_dev_id_[c_]string()
[2] libata: reorganize ata_dev_identify() (7 patches)
[3] libata: implement and use ata_dev_revalidate() (4 patches)

Once these revalidation changes make into tree, I'll submit ata_set_mode() changes (I've submitted the first patchset of this but Alan had a valid objection. I agreed and dropped that one.[4])

After ata_set_mode() changes are complete, the real EH changes will be submitted including error recording (to determine whether speed down is necessary) and all the recovery actions. I haven't splitted these yet but I guess it will be three or four patchsets.

And after that, Jens' NCQ work can be integrated over it. The code required to implement NCQ is actually quite small. It will probably take only one patchset of about three to five patches.

So, this is where we stand. I hope we can get new EH up and working soon.

Thanks a lot. :-)

Signature · 6 lines

tejun

[1] http://article.gmane.org/gmane.linux.ide/7997
[2] http://article.gmane.org/gmane.linux.ide/8048
[3] http://article.gmane.org/gmane.linux.ide/8050
[4] http://article.gmane.org/gmane.linux.ide/8054

Albert Lee

2006-02-20 08:12 UTC #1 lore raw

Hi Tejun:

Quoted 9 lines

I think we're half-way through the EH changes. New reset mechanism and most of config reorganization are in tree. Currently, the patches which are waiting for your review are...
[1] libata: rename ata_dev_id_[c_]string()
[2] libata: reorganize ata_dev_identify() (7 patches)
[3] libata: implement and use ata_dev_revalidate() (4 patches)

Quoted 7 lines

Once these revalidation changes make into tree, I'll submit ata_set_mode() changes (I've submitted the first patchset of this but Alan had a valid objection. I agreed and dropped that one.[4])
After ata_set_mode() changes are complete, the real EH changes will be submitted including error recording (to determine whether speed down is necessary) and all the recovery actions.

Will PATA speed down be supported?
e.g. udma133 => udma100 => udma66 => udma33 => pio ...

Quoted 10 lines

I haven't splitted these yet
but I guess it will be three or four patchsets.
And after that, Jens' NCQ work can be integrated over it. The code required to implement NCQ is actually quite small. It will probably take only one patchset of about three to five patches.
So, this is where we stand. I hope we can get new EH up and working soon.

Albert

Tejun Heo

author

2006-02-20 08:19 UTC #2 lore raw

Quoted 14 lines · Albert Lee

Hi Tejun:
Quoted 12 lines
I think we're half-way through the EH changes. New reset mechanism and most of config reorganization are in tree. Currently, the patches which are waiting for your review are...
[1] libata: rename ata_dev_id_[c_]string()
[2] libata: reorganize ata_dev_identify() (7 patches)
[3] libata: implement and use ata_dev_revalidate() (4 patches)
Once these revalidation changes make into tree, I'll submit ata_set_mode() changes (I've submitted the first patchset of this but Alan had a valid objection. I agreed and dropped that one.[4])

Quoted 8 lines

Quoted 3 lines
After ata_set_mode() changes are complete, the real EH changes will be submitted including error recording (to determine whether speed down is necessary) and all the recovery actions.
Will PATA speed down be supported?
e.g. udma133 => udma100 => udma66 => udma33 => pio ...

Yeap, sure.

Signature · 1 line

tejun

Albert Lee

2006-02-20 08:27 UTC #3 lore raw

Quoted 22 lines · Tejun Heo

Quoted 20 lines · Albert Lee
Hi Tejun:
Quoted 15 lines
I think we're half-way through the EH changes. New reset mechanism and most of config reorganization are in tree. Currently, the patches which are waiting for your review are...
[1] libata: rename ata_dev_id_[c_]string()
[2] libata: reorganize ata_dev_identify() (7 patches)
[3] libata: implement and use ata_dev_revalidate() (4 patches)
Once these revalidation changes make into tree, I'll submit ata_set_mode() changes (I've submitted the first patchset of this but Alan had a valid objection. I agreed and dropped that one.[4])
After ata_set_mode() changes are complete, the real EH changes will be submitted including error recording (to determine whether speed down is necessary) and all the recovery actions.

Quoted 6 lines

Quoted 3 lines
Will PATA speed down be supported?
e.g. udma133 => udma100 => udma66 => udma33 => pio ...
Yeap, sure.

That's great, thanks! :)

Albert