diff options
| author | Arthur Hanson Rasmusson <arthur@vgpu.io> | 2026-10-23 20:01:53 -0400 |
|---|---|---|
| committer | Arthur Hanson Rasmusson <arthur@vgpu.io> | 2026-10-23 20:01:53 -0400 |
| commit | ccdc1240d5df9777503a05b247109b6136d03313 (patch) | |
| tree | 74f1ea2bede53a4d7c863ede4f861bc611a038ba /kernel/gsp_rm_emulation.h | |
| parent | 1c596b4b30e7ecc2f525ae47b52a73b8d4de816d (diff) | |
| download | open-iov-ccdc1240d5df9777503a05b247109b6136d03313.tar.gz open-iov-ccdc1240d5df9777503a05b247109b6136d03313.zip | |
vgsp-sema r577: never let a foreign release VA claim a CE channel's tracking slot; event-record A/B knobs; GR-DEC dump deepened
r575 starved-release wedge (live): an unreadable UVM CE entry's rescued
stale page decoded a release to a foreign VA (0x12006c004, pay=60); the
rescue paths adopted it as the channel's tracking slot (last_rel_va) and
every later recovery release targeted it, while the client's real pool
(0x121000000) stopped at 25 - the client spun in uvm_tracker_wait for 34
(exact-slot poke delivered it and the client completed green). For CE
channels only a pool-shaped release target (0x121000000..0x122000000) may
become last_rel_va; a foreign rescued release is refused (witnessed,
bounded) and the pool step is delivered through the exact-slot fallback /
deferred-release machinery instead. The fake's read-before-write keeps the
delivery monotonic even when a legit push carries multiple releases.
Also: gsp_rm_ev_{seq,nlist,cl} runtime knobs for the guest-RM event drop
A/B (the r576 experiment: guest consumes every POST_EVENT record and never
enters nv_post_event), and the GR-DEC launch dump now covers 256 data
pages (the mail VADD fill carries n_data=969).
New: tools/cuda-ab/evhold.cu - deterministic CUDA event holder/wait probe
(context + cudaEventSynchronize + a hold window) for the event experiments.
Diffstat (limited to 'kernel/gsp_rm_emulation.h')
| -rw-r--r-- | kernel/gsp_rm_emulation.h | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/kernel/gsp_rm_emulation.h b/kernel/gsp_rm_emulation.h index 9dc354d..85ee4d9 100644 --- a/kernel/gsp_rm_emulation.h +++ b/kernel/gsp_rm_emulation.h @@ -216,6 +216,15 @@ struct gsp_rm_emu { */ extern unsigned gsp_rm_ev_log_max; +/* + * r577 experiment knobs (runtime-writable; default 0 = legacy). See the + * definitions in gsp_rm_emulation.c: event record sequence, bNotifyList, + * hClient override - the A/B space for the guest-RM event drop. + */ +extern unsigned gsp_rm_ev_seq; +extern unsigned gsp_rm_ev_nlist; +extern unsigned gsp_rm_ev_cl; + int gsp_rm_emu_init(struct nv_vgpu_device *vgpu, struct gsp_emu_context *ctx); void gsp_rm_emu_fini(struct gsp_emu_context *ctx); int gsp_rm_post_channel_event(struct nv_vgpu_device *vgpu, u32 h_src); |