Change temporary entrypoints to be lazily allocated by davidwrighton · Pull Request #101580 · dotnet/runtime

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton · 2024-04-25T23:44:58Z

This change moves temporary entrypoints to be allocated when they are needed instead of eagerly at type load time

As you can see, with a very small test case the wins are significant, and notably, reducing allocations in the stub heaps is nice as they are difficult to allocate efficiently as they require a bunch of memory mapping operations. In addition, with this change I'm removing the concept of Compact Entrypoints, as they are no longer significantly beneficial as of the advent of tiered compilation, and would be difficult to integrate with this change.

Baseline
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xf0000 (983040) bytes total.
HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total.
NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted.
PR
Loader Heap:
----------------------------------------
LowFrequencyHeap: Size: 0xef000 (978944) bytes total.
HighFrequencyHeap: Size: 0x1ad000 (1757184) bytes total, 0x3000 (12288) bytes wasted.
StubHeap: Size: 0x1000 (4096) bytes total.
FixupPrecodeHeap: Size: 0x38000 (229376) bytes total.
NewStubPrecodeHeap: Size: 0x8000 (32768) bytes total.
IndirectionCellHeap: Size: 0x1000 (4096) bytes total.
CacheEntryHeap: Size: 0x1000 (4096) bytes total.
Total size: Size: 0x2df000 (3010560) bytes total, 0x3000 (12288) bytes wasted.

What can be seen here, is that the size of the Precode heaps shrinks dramatically, and the size of the HighFrequency heap grows a little bit. With a larger test case (launching powershell with a -c exit command line argument, the gains are still very similar. Also with powershell, I have a simple benchmark which runs pwsh.exe -c exit 100 times and measures the total execution time. The performance improvement is around 1% which I consider to be quite a nice result.

Architecture/Version	Total Size	Process Execution Time
Baseline X86	9,420,800 bytes	23.33 seconds
PR X86	6,348,800 bytes	23.18 seconds
Baseline X64	13,586,432 bytes	23.23 seconds
PR X64	10,440,704 bytes	23.15 seconds

Baseline Loader Heap: ---------------------------------------- System Domain: 7ffab916ec00 LoaderAllocator: 7ffab916ec00 LowFrequencyHeap: Size: 0xf0000 (983040) bytes total. HighFrequencyHeap: Size: 0x16a000 (1482752) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x168000 (1474560) bytes total. NewStubPrecodeHeap: Size: 0x18000 (98304) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x3dd000 (4050944) bytes total, 0x3000 (12288) bytes wasted. Compare Loader Heap: ---------------------------------------- System Domain: 7ff9eb49dc00 LoaderAllocator: 7ff9eb49dc00 LowFrequencyHeap: Size: 0xef000 (978944) bytes total. HighFrequencyHeap: Size: 0x1b2000 (1777664) bytes total, 0x3000 (12288) bytes wasted. StubHeap: Size: 0x1000 (4096) bytes total. FixupPrecodeHeap: Size: 0x70000 (458752) bytes total. NewStubPrecodeHeap: Size: 0x10000 (65536) bytes total. IndirectionCellHeap: Size: 0x1000 (4096) bytes total. CacheEntryHeap: Size: 0x1000 (4096) bytes total. Total size: Size: 0x324000 (3293184) bytes total, 0x3000 (12288) bytes wasted. LowFrequencyHeap is 4KB bigger HighFrequencyHeap is 288KB bigger FixupPrecodeHeap is 992KB smaller NewstubPrecodeHeap is 32KB smaller

…y definition the method is defining the slot

…ests

…y_entrypoints

… an interface

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…y_entrypoints

Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

AaronRobinsonMSFT · 2024-06-29T03:03:14Z

 if (m_pRepresentativeMT != NULL)
 {
- pMD = m_pRepresentativeMT->GetMethodDescForSlot(m_representativeSlot);
+ pMD = m_pRepresentativeMT->GetMethodDescForSlot_NoThrow(m_representativeSlot);


My point is if this is not supposed to return a NULL and we're calling the "NoThrow" then at the very least we need an assert to convey that assumption.

…sts...

Co-authored-by: Aaron Robinson <arobins@microsoft.com>

…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

…y_entrypoints

AaronRobinsonMSFT · 2024-07-12T20:39:45Z

+ if (RequiresStableEntryPoint())
+ {
+ GetOrCreatePrecode()->SetTargetInterlocked(entryPoint);
+ }
+ else
+ {
+ SetStableEntryPointInterlocked(entryPoint);
+ }


This code reads odd. If RequiresStableEntryPoint() returns false, then we set the stable entry point? If this is correct, I think a comment is appropriate.

Sigh. It is right, but yeah, this should be better structured. A RequiredStableEntryPoint is currently required have a stable entrypoint which is a Precode. Otherwise, we can generate set a stable entrypoint which isn't based on a Precode. Before this change, we would have reliably hit the HasPrecode flag above, but with making Precode generation lazy, this is a point where the Precode needs to be forced at this point. However, I think I'll fix this by changing the previous condition to be (RequiresStableEntryPoint()) and not use GetPrecode, and instead use GetOrCreatePrecode(). Then this portion of the condition can revert to be what it was before.

…sn't REQUIRED. we need to handle that case.

- Support iterating methods from the new ISOSDacInterface15 - Re-plumb the existing logic as a local implementation of ISOSDacInterface15 on top of apis that have been present for a long time This is dependent on dotnet/runtime#101580

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

davidwrighton added 5 commits April 25, 2024 08:20

WorkingOnIt
ab85b04

If there isn't a parent methodtable and the slot matches... then it b…
693d837
…y definition the method is defining the slot

Fix a couple more issues found when running a subset of the coreclr t…
da774fb
…ests

Get X86 building again
bb59e29

ghost added the area-VM-coreclr label Apr 25, 2024

dotnet-policy-serviceBot assigned davidwrighton Apr 25, 2024

davidwrighton added 3 commits May 1, 2024 16:05

Attempt to use a consistent api to force slots to be set
bfde6de

Put cache around RequiresStableEntryPoint
54b8ab6

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
7a4bd28
…y_entrypoints

This was referenced May 2, 2024
RestApiException`1: The response contained an invalid status code 500 Internal Server Error dotnet/dnceng#2298
Closed
System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanScalarDestination_SpecialValues fails #101721
Closed

davidwrighton added 3 commits May 3, 2024 15:17

Fix typo
66f2b39

Fix interop identified issue where we sometime set a non Precode into…
951a655
… an interface

Move ARM and X86 to disable compact entry points
45a0b3b

This was referenced May 7, 2024
System.IO.Net5Compat.Tests and System.IO.Tests suddenly exiting with error 137 #100558
Closed
Test failure in System.Numerics.Tensors.Tests.SingleGenericTensorPrimitives.SpanDestinationFunctions_SpecialValues #101731
Closed

davidwrighton added 2 commits May 7, 2024 13:09

Attempt to fix build breaks
4e7f41d

fix typo
f6e2fed

build-analysisBot mentioned this pull request May 9, 2024
arm32 fails in CI with "/lib/arm-linux-gnueabihf/libc.so.6: version `GLIBC_2.34' not found" #102030
Closed

Fix another Musl validation issue
f9de777

build-analysisBot mentioned this pull request May 9, 2024
Dead lettering tests #101524
Closed

davidwrighton added 3 commits May 10, 2024 11:00

More tweaks around NULL handling
730fd7c

Hopefully the last NULL issue
fb335e7

Fix more NULL issues
d31ebbb

davidwrightonand others added 4 commits June 28, 2024 11:59

Update src/coreclr/vm/methodtable.inl
5bf0fb3
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Update src/coreclr/vm/methodtable.h
7010052
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
d55c93f
…y_entrypoints

Fix GetMethodDescForSlot_NoThrow
fb2f987
Try removing EnsureSlotFilled Implement IsEligibleForTieredCompilation in terms of IsEligibleForTieredCompilation_NoCheckMethodDescChunk

build-analysisBot mentioned this pull request Jun 29, 2024
[mono][interpreter] Mono interpreter is crashing during System.Data.Odbc.Tests (linux-x64 Release Mono_Interpreter_LibrariesTests) #101370
Closed

AaronRobinsonMSFT reviewed Jun 29, 2024
View reviewed changes

davidwrighton added 8 commits July 1, 2024 11:45

Fix missing change intended in last commit
c5af140

Fix some more IsPublished memory use issues
0281127

Call the right GetSlot method
56137a2

Move another scenario to NoThrow, I think this should clear up our te…
98aa3c9
…sts...

Add additional IsPublished check
7cf8acd

Fix MUSL validation build error and Windows x86 build error
efe8f1a

Address code review feedback
d7d3948

Fix classcompat build
29571eb

build-analysisBot mentioned this pull request Jul 4, 2024
[NativeAOT] Object synchronization method was called from an unsynchronized block of code. #104340
Closed

Update src/coreclr/vm/method.cpp
3ce4e4a
Co-authored-by: Aaron Robinson <arobins@microsoft.com>

build-analysisBot mentioned this pull request Jul 5, 2024
Test failure: GC\\Features\\HeapExpansion\\Finalizer\\Finalizer.cmd #102706
Closed

davidwrighton added 2 commits July 9, 2024 09:59

Remove assert that is invalid because TryGetMulticCallableAddrOfCode …
a4c8803
…can return NULL ... and then another thread could produce a stable entrypoint and the assert could lose the race

Merge branch 'main' of github.com:dotnet/runtime into change_temporar…
99ddbcb
…y_entrypoints

build-analysisBot mentioned this pull request Jul 9, 2024
TimeProviderTests.TestProviderTimer failed in CI #103459
Closed

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

Comment threadsrc/coreclr/vm/method.cpp

AaronRobinsonMSFT reviewed Jul 12, 2024
View reviewed changes

AaronRobinsonMSFT approved these changes Jul 12, 2024
View reviewed changes

davidwrighton added 2 commits July 12, 2024 16:27

Final (hopefully) code review tweaks.
34e9a75

Its possible for GetOrCreatePrecode to be called for cases where it i…
2ed2b9c
…sn't REQUIRED. we need to handle that case.

davidwrighton merged commit dacf9db into dotnet:mainJul 14, 2024

jkotas mentioned this pull request Jul 16, 2024
Fix crash in DAC enum memory code that brings down createdump #104939
Merged

AndyAyersMS mentioned this pull request Jul 16, 2024
[Perf] Linux/x64: 6 Improvements on 7/14/2024 11:12:54 PM dotnet/perf-autofiling-issues#38285
Closed

github-actionsBot locked and limited conversation to collaborators Aug 14, 2024

Uh oh!

Conversation

davidwrighton commented Apr 25, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

davidwrighton commented Apr 25, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

davidwrighton commented Apr 25, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

davidwrighton commented Apr 25, 2024•
edited
Loading

davidwrighton commented Apr 25, 2024•
edited
Loading

davidwrighton commented Apr 25, 2024•
edited
Loading

davidwrighton commented Apr 25, 2024•
edited
Loading