quiche: update QUICHE to commit 5dd7a030209f9a6b5043bebd8ac3ee54f18d1d08#17328
quiche: update QUICHE to commit 5dd7a030209f9a6b5043bebd8ac3ee54f18d1d08#17328alyssawilk merged 14 commits intoenvoyproxy:mainfrom
Conversation
Signed-off-by: Dan Zhang <danzh@google.com>
Signed-off-by: Dan Zhang <danzh@google.com>
|
You have some CI issues but... /lgtm deps |
Signed-off-by: Dan Zhang <danzh@google.com>
|
/assign @wu-bin |
Signed-off-by: Dan Zhang <danzh@google.com>
Signed-off-by: Dan Zhang <danzh@google.com>
Signed-off-by: Dan Zhang <danzh@google.com>
Signed-off-by: Dan Zhang <danzh@google.com>
|
It's ready for review now. PTAL! |
| Envoy::Buffer::Instance* buffer_{nullptr}; | ||
| // Nullptr if this span is not constructed with a QuicMemSlice. | ||
| QuicMemSliceImpl* mem_slice_{nullptr}; |
There was a problem hiding this comment.
Can you add some comments to explain the relationship of buffer_ and mem_slice_? (It seems we only need one of them to be !nullptr at a time, which is easier to understand)
There was a problem hiding this comment.
done. buffer_ always points to the actually memory, but mem_slice_ is only used in the constructor which takes QuicMemSlice.
| const std::string bug_name_; | ||
| }; | ||
|
|
||
| class ScopedDisableExitOnQuicheBug { |
There was a problem hiding this comment.
nit: Clarify (in comments) it is test only, and only one instance can exists at a time.
There was a problem hiding this comment.
done. But why only one instance can exists at a time?
There was a problem hiding this comment.
I don't see comment added?
Suppose two threads, each has a ScopedDisableExitOnQuicheBug instance, called S1 and S2, and S1 and S2's lifetime overlaps like:
T=0: S1 construct
T=1: S2 construct
T=2: S1 destruct
T=3: S2 destruct
The final value for ScopedDisableExitOnQuicheBug will be true.
There was a problem hiding this comment.
Oh, you mean multiple thread. Add comment for real now.
There was a problem hiding this comment.
Yes multi thread is the main problem. But that sequence is problematic in single thread too.
Still not seeing comments. LGTM assuming you'll add one:)
There was a problem hiding this comment.
How can scoped objects have interleaved life time?
| abort(); | ||
| } | ||
| #else | ||
| Envoy::Assert::invokeEnvoyBugFailureRecordActionForEnvoyBugMacroUseOnly(bug_name_.data()); |
There was a problem hiding this comment.
invokeEnvoyBugFailureRecordActionForEnvoyBugMacroUseOnly() take a char*
| EXPECT_QUIC_BUG(bug("bug one is expected"), "bug one"); | ||
| EXPECT_QUIC_BUG(bug("bug two is expected"), "bug two"); | ||
| #ifdef NDEBUG | ||
| // The 3rd triggering in release mode should be omitted. |
There was a problem hiding this comment.
"should not be logged". If there's a bug counter it should still be incremented.
Signed-off-by: Dan Zhang <danzh@google.com>
wu-bin
left a comment
There was a problem hiding this comment.
Thanks Dan. LGTM but please add a comment for ScopedDisableExitOnQuicheBug.
| const std::string bug_name_; | ||
| }; | ||
|
|
||
| class ScopedDisableExitOnQuicheBug { |
There was a problem hiding this comment.
Yes multi thread is the main problem. But that sequence is problematic in single thread too.
Still not seeing comments. LGTM assuming you'll add one:)
|
(tagging as waiting on a new hash per internal discussion) |
Signed-off-by: Dan Zhang <danzh@google.com>
Updated to a newer commit. PTAL |
bazel/repository_locations.bzl
Outdated
| urls = ["https://storage.googleapis.com/quiche-envoy-integration/{version}.tar.gz"], | ||
| use_category = ["dataplane_core"], | ||
| release_date = "2021-06-02", | ||
| release_date = "2021-07-13", |
There was a problem hiding this comment.
I think the date for this commit is 2021-07-16 as per https://quiche.googlesource.com/quiche/+/5dd7a030209f9a6b5043bebd8ac3ee54f18d1d08 but it's a little difficult to navigate googlesource.com. As this isn't a Github repo I don't think we check this programatically
|
/lgtm deps |
Signed-off-by: Dan Zhang <danzh@google.com>
alyssawilk
left a comment
There was a problem hiding this comment.
LGTM modulo outstanding comments
|
@moderation I think this needs one more LGTM? |
|
/lgtm deps |
…d08 (envoyproxy#17328) Commit Message: Modify QUIC_BUG implementation to log with rate limit like ENVOY_BUG in release mode. Fix a QuicMemSliceSpanImpl life time issue which was exposed by new QUICHE change Risk Level: low Testing: added unit tests for QUIC_BUG and QuicMemSliceSpanImpl Signed-off-by: Dan Zhang <danzh@google.com>
Signed-off-by: Dan Zhang danzh@google.com
Commit Message: Modify QUIC_BUG implementation to log with rate limit like ENVOY_BUG in release mode. Fix a QuicMemSliceSpanImpl life time issue which was exposed by new QUICHE change
Risk Level: low
Testing: added unit tests for QUIC_BUG and QuicMemSliceSpanImpl