| Age | Commit message (Collapse) | Author |
|---|
|
CL 693398 returned the error from reading a generation immediately, but
this is wrong -- a Sync event must be emitted to indicate the end of the
trace before reporting the error. This caused TestCrashWhileTracing
to fail because that test has a high likelihood of producing a truncated
trace, and it expects at least 2 Sync events. The truncated trace error
would be reported before the second Sync event, which is incorrect.
Fixes #75045.
Change-Id: Ia71592c4ec56a544afc85cdb7b575e143f80e048
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Reviewed-on: https://go-review.googlesource.com/c/go/+/696436
Reviewed-by: Carlos Amedee <carlos@golang.org>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
|
This change takes the EvEndOfGeneration event and promotes it to a real
event that appears in the trace.
This allows the trace parser to unambiguously identify truncated traces
vs. broken traces. It also makes a lot of the logic around parsing
simpler, because there's no more batch spilling necessary.
Fixes #73904.
Change-Id: I37c359b32b6b5f894825aafc02921adeaacf2595
Reviewed-on: https://go-review.googlesource.com/c/go/+/693398
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
|
Replace the per-generation EvEventBatch containing a lone EvFrequency
event with a per-generation EvEventBatch containing a EvSync header
followed by an EvFrequency and EvClockSnapshot event.
The new EvClockSnapshot event contains trace, mono and wall clock
snapshots taken in close time proximity. Ignoring minor resolution
differences, the trace and mono clock are the same on linux, but not on
windows (which still uses a TSC based trace clock).
Emit the new sync batch at the very beginning of every new generation
rather than the end to be in harmony with the internal/trace reader
which emits a sync event at the beginning of every generation as well
and guarantees monotonically increasing event timestamps.
Bump the version of the trace file format to 1.25 since this change is
not backwards compatible.
Update the internal/trace reader implementation to decode the new
events, but do not expose them to the public reader API yet. This is
done in the next CL.
For #69869
Change-Id: I5bfedccdd23dc0adaf2401ec0970cbcc32363393
Reviewed-on: https://go-review.googlesource.com/c/go/+/653575
Reviewed-by: David Chase <drchase@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
|
|
CL 648315 and CL 648195 fixed #71615 in the case where we fail to read
the next generation by emitting an extra sync event before returning an
error. But, it's possible we failed to even read the next spilled batch
when we read the first generation, and have been carrying the error from
trying to read a spilled batch since the last generation. In this case,
we don't emit a final sync event, meaning that there are still some
cases where #71615 happens.
This change emits the final sync event in this corner case. I believe
this is the final corner case. I could previously reproduce the issue
by running the test under stress2, but I can no longer reproduce any
failures after this change.
Fixes #71615, for real this time.
Change-Id: I10688a3c0e4b8327a95f31add365338c77c091ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/649259
Reviewed-by: Cherry Mui <cherryyz@google.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
|
CL 648195 was supposed to have fixed #71615, but it didn't include an
update to r.syncs. I can confirm this CL fixes the issue even when
running the test many times in a row.
Fixes #71615.
Change-Id: I97db3d639dc5bc8648a191696f90b0e5087307c8
Reviewed-on: https://go-review.googlesource.com/c/go/+/648315
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Bypass: Michael Knyszek <mknyszek@google.com>
|
|
Since CL 644215 each Sync event now represents the coming generation,
with a final Sync event emitted even when there's nothing ahead. This
change however failed to emit a Sync event at the end of a completely
valid generation when the next generation was invalid, causing the
runtime test TestCrashWhileTracing to start failing.
Fix this by emitting a final Sync event even when the next generation is
broken. We hold onto the error in parsing the next generation and emit
it after that final Sync event.
(Should these "final" Sync events distinguish themselves in some way?)
Fixes #71615.
Change-Id: I1f8abee5abaa39e1219e6fa05e9f82f1478db4c9
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Reviewed-on: https://go-review.googlesource.com/c/go/+/648195
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
|
|
Currently all v2 trace versions, Go 1.22 and Go 1.23, share a full set
of specs. This is mostly OK, but it means quite a few events will be
accepted for 1.22 traces that should be rejected. This change fixes that
by limiting which event specs are returned by version.Version.Specs for
Go 1.22.
While we're here, let's be stricter about event names too, and move
tracev2.EventString to be a method on the version, so we can be more
precise. An intended consequence of this move is that tracev2 no longer
depends on fmt, since we will want the runtime to depend on tracev2 in
the near future.
Change-Id: If7285460c8ba59ab73da00993b7b12e61cdfe6a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/644219
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
|
|
This change follows up from the previous one which renamed oldtrace to
tracev1, defining everything Go 1.22+ as trace v2.
This change also re-maps some packages in preparation for sharing with
other parts of the standard library, like the runtime. It also cleans up
some other uses of 'go122' that are just a bit misleading. The mappings
are as follows:
- internal/trace/event -> internal/trace/tracev2/event
- internal/trace/event/go122 -> internal/trace/tracev2
- internal/trace/internal/testgen/go122 ->
internal/trace/internal/testgen
The CL updates all import paths and runs gofmt -w -s on the entire
subdirectory.
Change-Id: I35476c679a96d4eafad6b94bac5f88aa7b085d2f
Reviewed-on: https://go-review.googlesource.com/c/go/+/644218
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
|
This is part of a refactoring to better distinguish trace wire format
versions. Even though details may change between Go versions and they
might be backwards-incompatible, the trace format still broadly has two
wire formats: v1 and v2.
A follow-up change will rename go122 to v2 to make this more consistent.
Change-Id: If4fe1c82d8aeabc8baa05f525e08a9e7d469a5c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/644217
Auto-Submit: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
|
|
This change modifies how per-generation experimental batches are
exposed. Rather than expose them on the ExperimentalEvent, it exposes it
as part of the Sync event, so it's clear to the caller when the
information becomes relevant and when it should be parsed.
This change also adds a field to each ExperimentalEvent indicating which
experiment the event is a part of.
Because this information needs to appear *before* a generation is
observed, we now ensure there is a sync event both before and after each
generation. This means the final sync event is now a special case;
previously we would only emit a sync event after each generation.
This change is based on feedback from Austin Clements on the
experimental events functionality.
For #62627.
Change-Id: I48b0fe12b22abb7ac8820a9e73447bfed8419856
Reviewed-on: https://go-review.googlesource.com/c/go/+/644215
Auto-Submit: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
|
Multiple Ms can offer Events with identical timestamps. The Reader
edits those so the timestamps are strictly increasing, but it needs a
way to break the tie. Use something deterministic (such as the order of
the batches), rather than map iteration order.
Updates #68277
Change-Id: I4a1f70c1669ce1c9b52d09e2bc99acbc831ef9a4
Reviewed-on: https://go-review.googlesource.com/c/go/+/596355
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Rhys Hiltner <rhys.hiltner@gmail.com>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
|
|
This change moves the v2 tracer into the trace directory.
Updates #67367
Change-Id: I3657b4227002cb00fdf29c797434800ea796715e
Reviewed-on: https://go-review.googlesource.com/c/go/+/584538
Reviewed-by: Michael Knyszek <mknyszek@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
|
