PoC: avoid allocating GCHandle for ReceiveFromAsync on Windows #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
wfurt
pushed a commit
that referenced
this pull request
Nov 29, 2023
Fixes dotnet#95367. Relevant part of the JitDump: ``` Using `if true` assertions from pred BB02 Assertions in: #1 fgMorphTree BB04, STMT00021 (before) [000070] DA--------- * STORE_LCL_VAR ubyte V10 tmp9 [000057] ----------- \--* CAST int <- ubyte <- int [000006] ----------- \--* EQ int [000004] ----------- +--* LCL_VAR ref V02 tmp1 (last use) [000055] H---------- \--* CNS_INT(h) ref 'Frozen EmptyPartition`1<Int32> object' Assertion prop for index #1 in BB04: [000006] ----------- * EQ int GenTreeNode creates assertion: [000070] DA---+----- * STORE_LCL_VAR ubyte V10 tmp9 In BB04 New Local Constant Assertion: V10 == [0000000000000001], index = #2 fgMorphTree BB04, STMT00021 (after) [000070] DA---+----- * STORE_LCL_VAR ubyte V10 tmp9 [000055] H----+----- \--* CNS_INT(h) int ``` The JitDump is unfinished because the compiler crashes when trying to dump the last line. Clearly, the `CNS_INT` is no longer a handle at that point because we just bashed it to a constant 1.
wfurt
pushed a commit
that referenced
this pull request
Mar 17, 2024
CodeQL flagged various places where we're dereferencing pointers that could be NULL, this PR systematically cleans some of them up via g_assert. * g_assert result of g_build_path calls * Allocation failure handling * mono_class_inflate_generic_class_checked can return NULL
wfurt
pushed a commit
that referenced
this pull request
May 15, 2024
…#102133) This generalizes the indir reordering optimization (that currently only triggers for loads) to kick in for GT_STOREIND nodes. The main complication with doing this is the fact that the data node of the second indirection needs its own reordering with the previous indirection. The existing logic works by reordering all nodes between the first and second indirection that are unrelated to the second indirection's computation to happen after it. Once that is done we know that there are no uses of the first indirection's result between it and the second indirection, so after doing the necessary interference checks we can safely move the previous indirection to happen after the data node of the second indirection. Example: ```csharp class Body { public double x, y, z, vx, vy, vz, mass; } static void Advance(double dt, Body[] bodies) { foreach (Body b in bodies) { b.x += dt * b.vx; b.y += dt * b.vy; b.z += dt * b.vz; } } ``` Diff: ```diff @@ -1,18 +1,17 @@ -G_M55007_IG04: ;; offset=0x001C +G_M55007_IG04: ;; offset=0x0020 ldr x3, [x0, w1, UXTW #3] ldp d16, d17, [x3, #0x08] ldp d18, d19, [x3, #0x20] fmul d18, d0, d18 fadd d16, d16, d18 - str d16, [x3, #0x08] - fmul d16, d0, d19 - fadd d16, d17, d16 - str d16, [x3, #0x10] + fmul d18, d0, d19 + fadd d17, d17, d18 + stp d16, d17, [x3, #0x08] ldr d16, [x3, #0x18] ldr d17, [x3, #0x30] fmul d17, d0, d17 fadd d16, d16, d17 str d16, [x3, #0x18] add w1, w1, #1 cmp w2, w1 bgt G_M55007_IG04 ```
wfurt
pushed a commit
that referenced
this pull request
Sep 10, 2024
* bug #1: don't allow for values out of the SerializationRecordType enum range * bug #2: throw SerializationException rather than KeyNotFoundException when the referenced record is missing or it points to a record of different type * bug #3: throw SerializationException rather than FormatException when it's being thrown by BinaryReader (or sth else that we use) * bug #4: document the fact that IOException can be thrown * bug #5: throw SerializationException rather than OverflowException when parsing the decimal fails * bug #6: 0 and 17 are illegal values for PrimitiveType enum * bug #7: throw SerializationException when a surrogate character is read (so far an ArgumentException was thrown)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Because we allocate new SocketAddress on each call, following check is never true
https://github.com/dotnet/runtime/blob/849ed0a0b895edb6dc70c4521f2284af640ec9d6/src/libraries/System.Net.Sockets/src/System/Net/Sockets/SocketAsyncEventArgs.Windows.cs#L863-L864
The fix is to use NativeMemory that would persist across all calls (for given AddressFamily)
The downside is that we need to copy the address but that seems small cost comparing to GCHandle allocation and management.
closes dotnet#86513
contributes to dotnet#30797