libvmmapi: Make memory segment handling a bit more abstract
ClosedPublic
Actions

Authored by markj on Jul 12 2023, 1:53 PM.

Details

Reviewers

jhb
corvink
andrew

Group Reviewers

bhyve

Commits

rG7e0fa794123e: libvmmapi: Make memory segment handling a bit more abstract

Summary

libvmmapi leaves a hole at [3GB, 4GB) in the guest physical address
space. This hole is not used in the arm64 port, which maps everything
above 4GB. This change makes the code a bit more general to accomodate
arm64 more naturally. In particular:

Remove vm_set_lowmem_limit(): it is unused and doesn't have well-defined constraints, e.g., nothing prevents a consumer from setting a lowmem limit above the highmem base.
Define a constant for the highmem base and use that everywhere that the base is currently hard-coded.
Make the lowmem limit a compile-time constant instead of a vmctx field.
Store segment info in an array.
Add vm_get_highmem_base(), for use in bhyve since the current value is hard-coded in some places.

No functional change intended.

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Build Status

Buildable 52597
Build 49488: arc lint + arc unit

Event Timeline

markj created this revision.Jul 12 2023, 1:53 PM

Herald added a reviewer: bhyve. · View Herald TranscriptJul 12 2023, 1:53 PM

Herald added subscribers: rgrimes, imp. · View Herald Transcript

markj requested review of this revision.Jul 12 2023, 1:53 PM

Harbormaster completed remote builds in B52597: Diff 124566.Jul 12 2023, 1:53 PM

markj added a parent revision: D41003: libvmmapi: Move PCI passthrough ioctl wrappers into a separate file.Jul 12 2023, 1:53 PM

markj added a child revision: D41005: libvmmapi: Add arm64 support.

I do wonder if we want to make even the libvmmapi interface more abstract where the exposed APIs take a segment index. Ideally you'd not even call it highmem on arm64, but only care about lowmem on x86. I'm thinking a bit of how we do this in the kernel where use MD macros to enable certain freelists (where x86 has an extra freelist for the low 1MB IIRC)

lib/libvmmapi/vmmapi.c
449	This could also maybe be a for-loop?
482	Seems like this could be a for loop over the ctx->memsegs array now?

This revision is now accepted and ready to land.Jul 12 2023, 7:06 PM

corvink accepted this revision.Jul 13 2023, 6:56 AM

corvink added inline comments.

lib/libvmmapi/internal.h
23	Adding a limit would allow us to make this code more generic. We could create a MD array for the memsegs like we did for ioctls in libvmmapi.
lib/libvmmapi/vmmapi.c
425	This could be a for-loop too when adding a limit to memegs: size_t remaining = len; for (i = 0; i < nitems(ctx->memsegs); ++i) { if (remaining < ctx->memsegs[i].limit - ctx->memsegs[i].base) { ctx->memsegs[i].size = remaining; break; } ctx->memsegs[i].size = ctx->memsegs[i].limit - memsegs[i].base; remaining -= ctx->memsegs[i].size; }
506	This could use a for-loop too.
535	This functions seems to be unused.