Handle large mallocs by going directly to kmem without taking a detour through uma zones.
ClosedPublic
Actions

Authored by jeff on Nov 26 2019, 10:22 PM.

Details

Reviewers

markj
rlibby

Commits

rS355203: Handle large mallocs by going directly to kmem. Taking a detour through

Summary

In retrospect I don't know why i didn't do this in the first place. Rather than putting a slab in vtoslab() just so we can figure out what size the thing was, stuff the size and a special bit in the 'slab' pointer in the page. We can drop the slab allocation entirely.

I can do something similar for large zone sizes later.

Diff Detail

Lint

Lint Passed

Unit

No Test Coverage

Build Status

Buildable 27823
Build 26002: arc lint + arc unit

Event Timeline

jeff created this revision.Nov 26 2019, 10:22 PM

Harbormaster completed remote builds in B27789: Diff 64918.Nov 26 2019, 10:22 PM

jeff retitled this revision from Handle large mallocs by going directly to kmem without taking a detour through uma zones. to Handle large mallocs by going directly to kmem without taking a detourthrough uma zones..Nov 26 2019, 10:25 PM

jeff edited the summary of this revision. (Show Details)

jeff added reviewers: markj, rlibby.

I guess using UMA made more sense when KVA was still allocated directly from the vm_map instead of vmem?

sys/kern/kern_malloc.c
673	I don't think this behaviour from uma_large_malloc_domain() needs to be carried over. malloc_domain() is only called from malloc_domainset(); callers of malloc_domainset() must use _PREF to avoid failures caused by empty domains. Actually this is all a big strange: malloc_domainset() is extracting a domain from a policy, and here we're converting it back to a policy. I think malloc_large() can change to use kmem_malloc_domain() instead. Otherwise we are running two iterators at different layers.

jeff added inline comments.Nov 27 2019, 8:49 PM

sys/kern/kern_malloc.c
673	It is a bit silly but uma_zalloc_domain is not uma_zalloc_domainset(). We could make such a thing and then collapse this. Or I could detect that the allocation should go to large in malloc_domainset() and bypass it.

markj added inline comments.Nov 27 2019, 10:11 PM

sys/kern/kern_malloc.c
673	I don't quite understand your first sentence. It would make most sense to me for malloc_large() to call kmem_malloc_domain().

jeff retitled this revision from Handle large mallocs by going directly to kmem without taking a detourthrough uma zones. to Handle large mallocs by going directly to kmem without taking a detour through uma zones..Nov 27 2019, 11:01 PM

Handle domain iterators differently. Make sure large sizes are recorded
correctly in malloc stats. Eliminate false failures from domain specific
allocations.

Harbormaster completed remote builds in B27823: Diff 64982.Nov 27 2019, 11:02 PM

markj accepted this revision.Nov 28 2019, 7:26 PM

markj added inline comments.