Skip to contentSkip to conversation

fs_dax_get_by_host in the hot path

DISCUSSION Christoph Hellwig 2 participants 3 messages Aug 24, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi Dan,

I just noticed that we now do a fs_dax_get_by_host in every iomap_begin call for DAX. This function iterates a list, does a string compared and igrab. I really think we need to cache this in the superblock (possible even the fs superblock) similar to what we do for the block device.
_______________________________________________
Linux-nvdimm mailing list [email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm

#1 lore raw
Quoted 7 lines · Christoph Hellwig

Hi Dan,

I just noticed that we now do a fs_dax_get_by_host in every iomap_begin call for DAX. This function iterates a list, does a string compared and igrab. I really think we need to cache this in the superblock (possible even the fs superblock) similar to what we do for the block device.

Sure, I'll get that fixed up. I believe I arrived at this scheme as a side-effect / misunderstanding of this comment:

https://groups.google.com/d/msg/linux.kernel/9ho8SFe6RSw/6zTgNaNTCgAJ
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm

#2 lore raw
Quoted 4 lines · Dan Williams

Sure, I'll get that fixed up. I believe I arrived at this scheme as a side-effect / misunderstanding of this comment:

https://groups.google.com/d/msg/linux.kernel/9ho8SFe6RSw/6zTgNaNTCgAJ

The current scheme is doing the right thing in terms of returning the dax device from ->iomap_begin. But it still do so efficiently :)
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm