Enable identical code folding on Linux by MichalStrehovsky · Pull Request #87045 · dotnet/runtime

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky · 2023-06-02T05:49:38Z

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons: 1. We weren't passing the command line switch. 2. We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (dotnet/corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into `.text` section, so let's just do the same. I also found out why `--foldmethodbodies` doesn't work outside Windows - we place managed code into a `__managedcode` section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this `.managedcode` like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

ghost · 2023-06-02T05:49:42Z

Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas
See info in area-owners.md if you want to be subscribed.

Issue Details

This instructs the linker to deduplicate identical sections on Linux. We were previously not doing this for two reasons:

We weren't passing the command line switch.
We had trouble getting linker to actually fold things. Linux linkers are particularly picky on what they're willing to fold. We've been historically putting these things to RW sections (Move InterfaceDispatchMapTable to data section corert#686) and that doesn't help. Looking at C++, it looks like it places vtables into .text section, so let's just do the same.

I also found out why --foldmethodbodies doesn't work outside Windows - we place managed code into a __managedcode section and Linux linkers won't fold sections that have names that are valid C identifiers (would need to name this .managedcode like on Windows) - https://github.com/dotnet/llvm-project/blob/c01ca3bc8a420b3796d816c43bdd06e337c72ea6/lld/ELF/ICF.cpp#L192. Not addressing that part because the whole thing looks like a hairy yak and this option is not supported anyway.

Cc @dotnet/ilc-contrib

Author:	MichalStrehovsky
Assignees:	-
Labels:	`area-NativeAOT-coreclr`
Milestone:	-

MichalStrehovsky · 2023-06-02T05:50:43Z

After the second commit this is just an okay saving:

	LLD Baseline	LLD PR
DynamicGenerics	5,059,776	5,036,560
HardwareIntrinsics	1,608,008	1,600,968

jkotas · 2023-06-02T06:43:02Z

@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
 if (factory.Target.IsWindows)


Should we extract this piece of logic into a helper method on ObjectNodeSection or NodeFactory so that we do not need to fix up a bunch of places everytime a new right way to do foldable readonly data sections shows up?

filipnavara · 2023-06-02T07:19:28Z

Can we run the NativeAOT extras pipeline for this? There's lot of quirks with ld64 on macOS and I am bit worried that the section rearrangement may trip up something.

MichalStrehovsky · 2023-06-02T08:38:17Z

/azp run runtime-extra-platforms

azure-pipelines · 2023-06-02T08:38:38Z

Azure Pipelines successfully started running 1 pipeline(s).

am11 · 2023-06-03T11:56:59Z

 <LinkerArg Include="-Wl,-segprot,__THUNKS,rx,rx" Condition="'$(_IsiOSLikePlatform)' == 'true'" />
 <LinkerArg Include="@(NativeFramework->'-framework %(Identity)')" Condition="'$(_IsApplePlatform)' == 'true'" />
 <LinkerArg Include="-Wl,--eh-frame-hdr" Condition="'$(_IsApplePlatform)' != 'true'" />
+ <LinkerArg Include="-Wl,--icf=all" Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />


Suggested change
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />
as moldalso supports it.

Extracting a piece of dotnet#87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

Extracting a piece of #87045 that I had to revert in that PR. Native linkers don't like when LSDA is in a COMDAT so fold these in the object writer instead. Seems to save about 1.2% in the Stage1 app. Obviously Unix only.

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

MichalStrehovsky added the area-NativeAOT-coreclr label Jun 2, 2023

ghost assigned MichalStrehovsky Jun 2, 2023

Update ObjectWriter.cs
b3af9ff

jkotas reviewed Jun 2, 2023
View reviewed changes

build-analysisBot mentioned this pull request Jun 2, 2023
Failed USB connection via port 54050, error 61, in tvOS arm64 Release AllSubsets_Mono #82637
Closed

am11 reviewed Jun 3, 2023
View reviewed changes

am11 mentioned this pull request Oct 3, 2023
[CI ONLY] [NativeAOT] ObjWriter in C# #92705
Closed
7 tasks

MichalStrehovsky mentioned this pull request Feb 5, 2024
Deduplicate LSDA data in the object writer #97972
Merged

MichalStrehovsky closed this Dec 9, 2024

MichalStrehovsky deleted the icflinux branch December 9, 2024 07:51

github-actionsBot locked and limited conversation to collaborators Jan 8, 2025

		@@ -19,7 +19,7 @@ public override ObjectNodeSection GetSection(NodeFactory factory)
		if (factory.Target.IsWindows)

	<LinkerArgInclude="-Wl,--icf=all"Condition="'$(LinkerFlavor)' == 'lld' or '$(LinkerFlavor)' == 'gold'" />
	<LinkerArgInclude="-Wl,--icf=all"Condition="'$(_IsApplePlatform)' != 'true' and '$(LinkerFlavor)' != 'bfd'" />

Uh oh!

Conversation

MichalStrehovsky commented Jun 2, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ghost commented Jun 2, 2023

Uh oh!

MichalStrehovsky commented Jun 2, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jkotasJun 2, 2023

Choose a reason for hiding this comment

Uh oh!

filipnavara commented Jun 2, 2023

Uh oh!

MichalStrehovsky commented Jun 2, 2023

Uh oh!

azure-pipelinesBot commented Jun 2, 2023

Uh oh!

am11Jun 3, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

MichalStrehovsky commented Jun 2, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ghost commented Jun 2, 2023

Uh oh!

MichalStrehovsky commented Jun 2, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jkotasJun 2, 2023

Choose a reason for hiding this comment

Uh oh!

filipnavara commented Jun 2, 2023

Uh oh!

MichalStrehovsky commented Jun 2, 2023

Uh oh!

azure-pipelinesBot commented Jun 2, 2023

Uh oh!

am11Jun 3, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

MichalStrehovsky commented Jun 2, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ghost commented Jun 2, 2023

Uh oh!

MichalStrehovsky commented Jun 2, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jkotasJun 2, 2023

Choose a reason for hiding this comment

Uh oh!

filipnavara commented Jun 2, 2023

Uh oh!

MichalStrehovsky commented Jun 2, 2023

Uh oh!

azure-pipelinesBot commented Jun 2, 2023

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

am11Jun 3, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

am11Jun 3, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

am11Jun 3, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading

am11Jun 3, 2023•
edited
Loading

MichalStrehovsky commented Jun 2, 2023•
edited
Loading