summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Modify README to reflect current progressHEADmasterLouis Burda2023-02-091-5/+9
|
* Most consistent version of prime self-testsLouis Burda2023-02-093-9/+25
|
* Added more tests and sanity checksLouis Burda2023-02-097-125/+434
|
* Add perf-based test for irqs interrupting prime-probe in test/evictionLouis Burda2023-02-071-0/+5
|
* fixup! Add syscall to deinit events to prevent blocking on sendLouis Burda2023-02-071-2/+2
|
* Add counter for L1 misses in guestLouis Burda2023-02-079-12/+32
|
* Add syscall to deinit events to prevent blocking on sendLouis Burda2023-02-0710-52/+92
|
* Add combined testcase based on exit codesLouis Burda2023-02-077-84/+39
|
* fixup! Swap gp registers to xmm in vmenterLouis Burda2023-02-065-7/+10
|
* Add kvm-targetstep to gitignoreLouis Burda2023-02-064-1/+2
|
* Swap gp registers to xmm in vmenterLouis Burda2023-02-063-8/+0
|
* Add aes crypto stepping initial testLouis Burda2023-02-066-18/+258
|
* qemu-targetstep: Attempt to track guest process gfn when running in userspaceLouis Burda2023-02-0612-12/+60
| | | | Seems like single-stepping the guest with LAPIC influences the guest scheduler behaviour, since just a single step inside the target gfn (to determine if its running in userspace), is enough to for us to never reach the GUEST_STOP_TRACK event. FWICT the single-stepping is not frequent and does not take long enough to justify never reaching the stop event.
* Fix stepping inconsistency by moving oneshot after primeeLouis Burda2023-02-069-28/+40
|
* Simplify target gfn stepping by tracking a single pageLouis Burda2023-02-033-11/+2
|
* Properly implement target gfn steppingLouis Burda2023-02-0319-1151/+428
|
* qemu-pagestep is quasi-realtime when suppresing slow kernel loggingLouis Burda2023-02-029-41/+85
|
* Debug single-stepping on page-boundaries in page_stepLouis Burda2023-02-029-11/+19
|
* Set default frequency to 1.5GHz and adjust apic timer divide accordinglyLouis Burda2023-02-024-5/+9
|
* Add event batching and handling of instructions loads on page boundaries for ↵Louis Burda2023-02-0216-36/+117
| | | | CPC_TRACK_PAGES
* Add guest provisioning scripts and instructionsLouis Burda2023-02-019-42/+155
|
* fixup! Event batchingLouis Burda2023-02-016-24/+21
|
* Update READMELouis Burda2023-02-011-15/+29
|
* Properly initialize locksLouis Burda2023-02-013-1/+4
|
* Remove kernel config for previous version and updated READMELouis Burda2023-02-014-10360/+22
|
* Fix build, add more build instructions to READMELouis Burda2023-02-014-12/+19
|
* Add track config to TRACK_MODE ioctl and combine TRACK_STEPS_* modesLouis Burda2023-01-3012-575/+68
|
* Event batchingLouis Burda2023-01-305-83/+154
|
* Added mainpfn guest utility to determine rough pfn for userspace codeLouis Burda2023-01-305-10/+69
|
* Make CPC_TRACK_PAGES_RESOLVE the default since no extra overheadLouis Burda2023-01-279-31/+21
|
* Consistent use of cpc shorthand instead of cachepcLouis Burda2023-01-2711-447/+428
|
* Add CPC_TRACK_STEPS and make apic reset more robustLouis Burda2023-01-2711-20/+28
|
* Added initial qemu-eviction and qemu-pagestepLouis Burda2023-01-2716-48/+162
| | | | Qemu-eviction seems to get stuck somewhere, potentially in the VC-handler(?) since we use the active gfn after resuming execution. Added qemu-pagestep to show viability of page-stepping for later use.
* Basic qemu eviction test with signalled stepping and added rip syscallLouis Burda2023-01-2618-523/+303
|
* Use mfence instead of lfence for memory barrierLouis Burda2023-01-255-4/+7
|
* Debug copy_to_user occasionally failing in kvm-stepLouis Burda2023-01-256-33/+29
|
* Remove cachepc objects from kvm-amd moduleLouis Burda2023-01-257-36/+38
|
* Replace bloated cache_ds alloc methodLouis Burda2023-01-258-471/+87
|
* Create flat binaries to support more complex guestsLouis Burda2023-01-2420-124/+178
|
* Test page tracking with potential branch prediction in guestLouis Burda2023-01-241-0/+10
|
* Page track support (without speculation)Louis Burda2023-01-245-97/+62
|
* Move kvm to guest process and add ipc for synchronizationLouis Burda2023-01-239-639/+317
|
* Remove unused ioctlsLouis Burda2023-01-233-69/+1
|
* Specify which revisions are out-of-sync with kernel changesLouis Burda2023-01-231-3/+2
|
* Use 16-bit realmode assembly for guests (!)Louis Burda2023-01-2313-47/+200
|
* Reimplement test and prime+probe in asm, make self-tests harder and improve ↵Louis Burda2023-01-2115-611/+369
| | | | noise
* Stash asm version of hwpf testLouis Burda2023-01-194-21/+85
|
* Many fixes, more precise single-stepping and more robust self-testsLouis Burda2023-01-1910-367/+323
|
* Fixup kvm-eviction and refactor kvm-stepLouis Burda2023-01-1719-1535/+1096
|
* Remove outdated kernel patchLouis Burda2023-01-121-1185/+0
|
generated by cgit v1.2.3-71-gd317 (git 2.46.0) at 2025-12-05 13:08:38 +0000