* Peter Xu (pet...@redhat.com) wrote: > This is v2 of postcopy preempt series. It can also be found here: > > https://github.com/xzpeter/qemu/tree/postcopy-preempt > > RFC: > https://lore.kernel.org/qemu-devel/20220119080929.39485-1-pet...@redhat.com > V1: > https://lore.kernel.org/qemu-devel/20220216062809.57179-1-pet...@redhat.com
I've queued some of this: tests: Pass in MigrateStart** into test_migrate_start() migration: Add migration_incoming_transport_cleanup() migration: postcopy_pause_fault_thread() never fails migration: Enlarge postcopy recovery to capture !-EIO too migration: Move static var in ram_block_from_stream() into global migration: Add postcopy_thread_create() migration: Dump ramblock and offset too when non-same-page detected migration: Introduce postcopy channels on dest node migration: Tracepoint change in postcopy-run bottom half migration: Finer grained tracepoints for POSTCOPY_LISTEN migration: Dump sub-cmd name in loadvm_process_command tp > v1->v2 changelog: > - Picked up more r-bs from Dave > - Rename both fault threads to drop "qemu/" prefix [Dave] > - Further rework on postcopy recovery, to be able to detect qemufile errors > from either main channel or postcopy one [Dave] > - shutdown() qemufile before close on src postcopy channel when postcopy is > paused [Dave] > - In postcopy_preempt_new_channel(), explicitly set the new channel in > blocking state, even if it's the default [Dave] > - Make RAMState.postcopy_channel unsigned int [Dave] > - Added patches: > - "migration: Create the postcopy preempt channel asynchronously" > - "migration: Parameter x-postcopy-preempt-break-huge" > - "migration: Add helpers to detect TLS capability" > - "migration: Fail postcopy preempt with TLS" > - "tests: Pass in MigrateStart** into test_migrate_start()" > > Abstract > ======== > > This series added a new migration capability called "postcopy-preempt". It > can > be enabled when postcopy is enabled, and it'll simply (but greatly) speed up > postcopy page requests handling process. > > Below are some initial postcopy page request latency measurements after the > new series applied. > > For each page size, I measured page request latency for three cases: > > (a) Vanilla: the old postcopy > (b) Preempt no-break-huge: preempt enabled, > x-postcopy-preempt-break-huge=off > (c) Preempt full: preempt enabled, > x-postcopy-preempt-break-huge=on > (this is the default option when preempt > enabled) > > Here x-postcopy-preempt-break-huge parameter is just added in v2 so as to > conditionally disable the behavior to break sending a precopy huge page for > debugging purpose. So when it's off, postcopy will not preempt precopy > sending a huge page, but still postcopy will use its own channel. > > I tested it separately to give a rough idea on which part of the change > helped how much of it. The overall benefit should be the comparison > between case (a) and (c). > > |-----------+---------+-----------------------+--------------| > | Page size | Vanilla | Preempt no-break-huge | Preempt full | > |-----------+---------+-----------------------+--------------| > | 4K | 10.68 | N/A [*] | 0.57 | > | 2M | 10.58 | 5.49 | 5.02 | > | 1G | 2046.65 | 933.185 | 649.445 | > |-----------+---------+-----------------------+--------------| > [*]: This case is N/A because 4K page does not contain huge page at all > > [1] > https://github.com/xzpeter/small-stuffs/blob/master/tools/huge_vm/uffd-latency.bpf > > TODO List > ========= > > TLS support > ----------- > > I only noticed its missing very recently. Since soft freeze is coming, and > obviously I'm still growing this series, so I tend to have the existing > material discussed. Let's see if it can still catch the train for QEMU 7.0 > release (soft freeze on 2022-03-08).. > > Avoid precopy write() blocks postcopy > ------------------------------------- > > I didn't prove this, but I always think the write() syscalls being blocked > for precopy pages can affect postcopy services. If we can solve this > problem then my wild guess is we can further reduce the average page > latency. > > Two solutions at least in mind: (1) we could have made the write side of > the migration channel NON_BLOCK too, or (2) multi-threads on send side, > just like multifd, but we may use lock to protect which page to send too > (e.g., the core idea is we should _never_ rely anything on the main thread, > multifd has that dependency on queuing pages only on main thread). > > That can definitely be done and thought about later. > > Multi-channel for preemption threads > ------------------------------------ > > Currently the postcopy preempt feature use only one extra channel and one > extra thread on dest (no new thread on src QEMU). It should be mostly good > enough for major use cases, but when the postcopy queue is long enough > (e.g. hundreds of vCPUs faulted on different pages) logically we could > still observe more delays in average. Whether growing threads/channels can > solve it is debatable, but sounds worthwhile a try. That's yet another > thing we can think about after this patchset lands. > > Logically the design provides space for that - the receiving postcopy > preempt thread can understand all ram-layer migration protocol, and for > multi channel and multi threads we could simply grow that into multile > threads handling the same protocol (with multiple PostcopyTmpPage). The > source needs more thoughts on synchronizations, though, but it shouldn't > affect the whole protocol layer, so should be easy to keep compatible. > > Patch Layout > ============ > > Patch 1-3: Three leftover patches from patchset "[PATCH v3 0/8] migration: > Postcopy cleanup on ram disgard" that I picked up here too. > > https://lore.kernel.org/qemu-devel/20211224065000.97572-1-pet...@redhat.com/ > > migration: Dump sub-cmd name in loadvm_process_command tp > migration: Finer grained tracepoints for POSTCOPY_LISTEN > migration: Tracepoint change in postcopy-run bottom half > > Patch 4-9: Original postcopy preempt RFC preparation patches (with slight > modifications). > > migration: Introduce postcopy channels on dest node > migration: Dump ramblock and offset too when non-same-page detected > migration: Add postcopy_thread_create() > migration: Move static var in ram_block_from_stream() into global > migration: Add pss.postcopy_requested status > migration: Move migrate_allow_multifd and helpers into migration.c > > Patch 10-15: Some newly added patches when working on postcopy recovery > support. After these patches migrate-recover command will allow re-entrance, > which is a very nice side effect. > > migration: Enlarge postcopy recovery to capture !-EIO too > migration: postcopy_pause_fault_thread() never fails > migration: Export ram_load_postcopy() > migration: Move channel setup out of postcopy_try_recover() > migration: Add migration_incoming_transport_cleanup() > migration: Allow migrate-recover to run multiple times > > Patch 16-19: The major work of postcopy preemption implementation is split > into > four patches as suggested by Dave. > > migration: Add postcopy-preempt capability > migration: Postcopy preemption preparation on channel creation > migration: Postcopy preemption enablement > migration: Postcopy recover with preempt enabled > > Patch 20-23: Newly added patches in this v2 for different purposes. > Majorly some amendment on existing postcopy preempt. > > migration: Create the postcopy preempt channel asynchronously > migration: Parameter x-postcopy-preempt-break-huge > migration: Add helpers to detect TLS capability > migration: Fail postcopy preempt with TLS for now > > Patch 24-25: Test cases (including one more patch for cleanup) > > tests: Add postcopy preempt test > tests: Pass in MigrateStart** into test_migrate_start() > > Please review, thanks. > > Peter Xu (25): > migration: Dump sub-cmd name in loadvm_process_command tp > migration: Finer grained tracepoints for POSTCOPY_LISTEN > migration: Tracepoint change in postcopy-run bottom half > migration: Introduce postcopy channels on dest node > migration: Dump ramblock and offset too when non-same-page detected > migration: Add postcopy_thread_create() > migration: Move static var in ram_block_from_stream() into global > migration: Add pss.postcopy_requested status > migration: Move migrate_allow_multifd and helpers into migration.c > migration: Enlarge postcopy recovery to capture !-EIO too > migration: postcopy_pause_fault_thread() never fails > migration: Export ram_load_postcopy() > migration: Move channel setup out of postcopy_try_recover() > migration: Add migration_incoming_transport_cleanup() > migration: Allow migrate-recover to run multiple times > migration: Add postcopy-preempt capability > migration: Postcopy preemption preparation on channel creation > migration: Postcopy preemption enablement > migration: Postcopy recover with preempt enabled > migration: Create the postcopy preempt channel asynchronously > migration: Parameter x-postcopy-preempt-break-huge > migration: Add helpers to detect TLS capability > migration: Fail postcopy preempt with TLS for now > tests: Add postcopy preempt test > tests: Pass in MigrateStart** into test_migrate_start() > > migration/channel.c | 10 +- > migration/migration.c | 235 ++++++++++++++++++++----- > migration/migration.h | 98 ++++++++++- > migration/multifd.c | 26 +-- > migration/multifd.h | 2 - > migration/postcopy-ram.c | 244 +++++++++++++++++++++----- > migration/postcopy-ram.h | 15 ++ > migration/qemu-file.c | 27 +++ > migration/qemu-file.h | 1 + > migration/ram.c | 330 +++++++++++++++++++++++++++++++---- > migration/ram.h | 3 + > migration/savevm.c | 70 ++++++-- > migration/socket.c | 22 ++- > migration/socket.h | 1 + > migration/trace-events | 19 +- > qapi/migration.json | 8 +- > tests/qtest/migration-test.c | 68 ++++++-- > 17 files changed, 983 insertions(+), 196 deletions(-) > > -- > 2.32.0 > > -- Dr. David Alan Gilbert / dgilb...@redhat.com / Manchester, UK
