[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan created this revision.
nruslan added a reviewer: clang.
Herald added a subscriber: cfe-commits.

Adds support for this flag. There is also another piece for llvm (separate 
review). More info: https://bugs.llvm.org/show_bug.cgi?id=36221


Repository:
  rC Clang

https://reviews.llvm.org/D43108

Files:
  docs/ClangCommandLineReference.rst
  include/clang/Driver/Options.td
  include/clang/Frontend/CodeGenOptions.def
  lib/CodeGen/TargetInfo.cpp
  lib/Driver/ToolChains/Clang.cpp
  lib/Frontend/CompilerInvocation.cpp

Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -914,6 +914,8 @@
 Opts.StackProbeSize = StackProbeSize;
   }
 
+  Opts.NoStackArgProbe = Args.hasArg(OPT_mno_stack_arg_probe);
+
   if (Arg *A = Args.getLastArg(OPT_fobjc_dispatch_method_EQ)) {
 StringRef Name = A->getValue();
 unsigned Method = llvm::StringSwitch(Name)
Index: lib/Driver/ToolChains/Clang.cpp
===
--- lib/Driver/ToolChains/Clang.cpp
+++ lib/Driver/ToolChains/Clang.cpp
@@ -4034,6 +4034,9 @@
   CmdArgs.push_back("-mstack-probe-size=0");
   }
 
+  if (Args.hasArg(options::OPT_mno_stack_arg_probe))
+CmdArgs.push_back("-mno-stack-arg-probe");
+
   if (Arg *A = Args.getLastArg(options::OPT_mrestrict_it,
options::OPT_mno_restrict_it)) {
 if (A->getOption().matches(options::OPT_mrestrict_it)) {
Index: lib/CodeGen/TargetInfo.cpp
===
--- lib/CodeGen/TargetInfo.cpp
+++ lib/CodeGen/TargetInfo.cpp
@@ -2355,16 +2355,17 @@
   }
 };
 
-static void addStackProbeSizeTargetAttribute(const Decl *D,
- llvm::GlobalValue *GV,
- CodeGen::CodeGenModule &CGM) {
+static void addStackProbeParamsTargetAttribute(const Decl *D,
+   llvm::GlobalValue *GV,
+   CodeGen::CodeGenModule &CGM) {
   if (D && isa(D)) {
-if (CGM.getCodeGenOpts().StackProbeSize != 4096) {
-  llvm::Function *Fn = cast(GV);
+llvm::Function *Fn = cast(GV);
 
+if (CGM.getCodeGenOpts().StackProbeSize != 4096)
   Fn->addFnAttr("stack-probe-size",
 llvm::utostr(CGM.getCodeGenOpts().StackProbeSize));
-}
+if (CGM.getCodeGenOpts().NoStackArgProbe)
+  Fn->addFnAttr("no-stack-arg-probe");
   }
 }
 
@@ -2374,7 +2375,7 @@
   X86_32TargetCodeGenInfo::setTargetAttributes(D, GV, CGM, IsForDefinition);
   if (!IsForDefinition)
 return;
-  addStackProbeSizeTargetAttribute(D, GV, CGM);
+  addStackProbeParamsTargetAttribute(D, GV, CGM);
 }
 
 class WinX86_64TargetCodeGenInfo : public TargetCodeGenInfo {
@@ -2436,7 +2437,7 @@
 }
   }
 
-  addStackProbeSizeTargetAttribute(D, GV, CGM);
+  addStackProbeParamsTargetAttribute(D, GV, CGM);
 }
 }
 
@@ -5622,7 +5623,7 @@
   ARMTargetCodeGenInfo::setTargetAttributes(D, GV, CGM, IsForDefinition);
   if (!IsForDefinition)
 return;
-  addStackProbeSizeTargetAttribute(D, GV, CGM);
+  addStackProbeParamsTargetAttribute(D, GV, CGM);
 }
 }
 
Index: include/clang/Frontend/CodeGenOptions.def
===
--- include/clang/Frontend/CodeGenOptions.def
+++ include/clang/Frontend/CodeGenOptions.def
@@ -223,6 +223,7 @@
 ///< alignment, if not 0.
 VALUE_CODEGENOPT(StackProbeSize, 32, 4096) ///< Overrides default stack
///< probe size, even if 0.
+CODEGENOPT(NoStackArgProbe, 1, 0) ///< Set when -mno-stack-arg-probe is used
 CODEGENOPT(DebugColumnInfo, 1, 0) ///< Whether or not to use column information
   ///< in debug info.
 
Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1822,6 +1822,8 @@
   HelpText<"Set the stack alignment">;
 def mstack_probe_size : Joined<["-"], "mstack-probe-size=">, Group, Flags<[CC1Option]>,
   HelpText<"Set the stack probe size">;
+def mno_stack_arg_probe : Flag<["-"], "mno-stack-arg-probe">, Group, Flags<[CC1Option]>,
+  HelpText<"Disable stack probes">;
 def mthread_model : Separate<["-"], "mthread-model">, Group, Flags<[CC1Option]>,
   HelpText<"The thread model to use, e.g. posix, single (posix by default)">, Values<"posix,single">;
 def meabi : Separate<["-"], "meabi">, Group, Flags<[CC1Option]>,
Index: docs/ClangCommandLineReference.rst
===
--- docs/ClangCommandLineReference.rst
+++ docs/ClangCommandLineReference.rst
@@ -2180,6 +2180,10 @@
 
 Set the stack probe size
 
+.. option:: -mno-stack-arg-probe
+
+Disable 

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@craig.topper, @MatzeB, @hans: Can someone take a look at this change?


Repository:
  rC Clang

https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@hans: One real-world example is when it is used to compile UEFI code using 
PE/COFF targets natively. Obviously, UEFI uses ABI which is basically almost 
the same as MS ABI, except that chkstk is not used. It mostly works (I actually 
was able to get it running) except the cases when the code contains 
variable-sized arrays allocated on stacks. Unfortunately, stack-probe-size will 
only help with fixed sized array but will not help to solve the problem 
described in the bug description since stack usage is unknown at compile time. 
MinGW does not have this problem because it provides this flag.

@MatzeB : there is a test on LLVM side (related review). Do you think the test 
is needed for clang side? If so, please let me know, what kind of test it is 
supposed to be.


Repository:
  rC Clang

https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan updated this revision to Diff 134377.
nruslan marked 3 inline comments as done.
nruslan added a comment.

Added mstack-arg-probe, tests, and changes related to other comments of the 
reviewers


https://reviews.llvm.org/D43108

Files:
  docs/ClangCommandLineReference.rst
  include/clang/Driver/Options.td
  include/clang/Frontend/CodeGenOptions.def
  lib/CodeGen/TargetInfo.cpp
  lib/Driver/ToolChains/Clang.cpp
  lib/Frontend/CompilerInvocation.cpp
  test/CodeGen/stack-arg-probe.c
  test/Driver/stack-arg-probe.c

Index: test/Driver/stack-arg-probe.c
===
--- /dev/null
+++ test/Driver/stack-arg-probe.c
@@ -0,0 +1,7 @@
+// RUN: %clang -### %s 2>&1 | FileCheck %s -check-prefix=STACKPROBE
+// RUN: %clang -### -mno-stack-arg-probe -mstack-arg-probe %s 2>&1 | FileCheck %s -check-prefix=STACKPROBE
+// RUN: %clang -### -mstack-arg-probe -mno-stack-arg-probe %s 2>&1 | FileCheck %s -check-prefix=NO-STACKPROBE
+// REQUIRES: clang-driver
+
+// NO-STACKPROBE: -mno-stack-arg-probe
+// STACKPROBE-NOT: -mno-stack-arg-probe
Index: test/CodeGen/stack-arg-probe.c
===
--- /dev/null
+++ test/CodeGen/stack-arg-probe.c
@@ -0,0 +1,8 @@
+// RUN: %clang_cc1 %s -triple=i686-windows-msvc -emit-llvm -o - -mno-stack-arg-probe | FileCheck %s -check-prefix=NO-STACKPROBE
+// RUN: %clang_cc1 %s -triple=i686-windows-msvc -emit-llvm -o - | FileCheck %s -check-prefix=STACKPROBE
+
+// NO-STACKPROBE: attributes #{{[0-9]+}} = {{{.*}} "no-stack-arg-probe"
+// STACKPROBE-NOT: attributes #{{[0-9]+}} = {{{.*}} "no-stack-arg-probe"
+
+void test1() {
+}
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -914,6 +914,8 @@
 Opts.StackProbeSize = StackProbeSize;
   }
 
+  Opts.NoStackArgProbe = Args.hasArg(OPT_mno_stack_arg_probe);
+
   if (Arg *A = Args.getLastArg(OPT_fobjc_dispatch_method_EQ)) {
 StringRef Name = A->getValue();
 unsigned Method = llvm::StringSwitch(Name)
Index: lib/Driver/ToolChains/Clang.cpp
===
--- lib/Driver/ToolChains/Clang.cpp
+++ lib/Driver/ToolChains/Clang.cpp
@@ -4034,6 +4034,10 @@
   CmdArgs.push_back("-mstack-probe-size=0");
   }
 
+  if (!Args.hasFlag(options::OPT_mstack_arg_probe,
+options::OPT_mno_stack_arg_probe, true))
+CmdArgs.push_back(Args.MakeArgString("-mno-stack-arg-probe"));
+
   if (Arg *A = Args.getLastArg(options::OPT_mrestrict_it,
options::OPT_mno_restrict_it)) {
 if (A->getOption().matches(options::OPT_mrestrict_it)) {
Index: lib/CodeGen/TargetInfo.cpp
===
--- lib/CodeGen/TargetInfo.cpp
+++ lib/CodeGen/TargetInfo.cpp
@@ -2355,16 +2355,15 @@
   }
 };
 
-static void addStackProbeSizeTargetAttribute(const Decl *D,
- llvm::GlobalValue *GV,
- CodeGen::CodeGenModule &CGM) {
-  if (D && isa(D)) {
-if (CGM.getCodeGenOpts().StackProbeSize != 4096) {
-  llvm::Function *Fn = cast(GV);
+static void addStackProbeTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
+  CodeGen::CodeGenModule &CGM) {
+  if (llvm::Function *Fn = dyn_cast_or_null(GV)) {
 
+if (CGM.getCodeGenOpts().StackProbeSize != 4096)
   Fn->addFnAttr("stack-probe-size",
 llvm::utostr(CGM.getCodeGenOpts().StackProbeSize));
-}
+if (CGM.getCodeGenOpts().NoStackArgProbe)
+  Fn->addFnAttr("no-stack-arg-probe");
   }
 }
 
@@ -2374,7 +2373,7 @@
   X86_32TargetCodeGenInfo::setTargetAttributes(D, GV, CGM, IsForDefinition);
   if (!IsForDefinition)
 return;
-  addStackProbeSizeTargetAttribute(D, GV, CGM);
+  addStackProbeTargetAttributes(D, GV, CGM);
 }
 
 class WinX86_64TargetCodeGenInfo : public TargetCodeGenInfo {
@@ -2436,7 +2435,7 @@
 }
   }
 
-  addStackProbeSizeTargetAttribute(D, GV, CGM);
+  addStackProbeTargetAttributes(D, GV, CGM);
 }
 }
 
@@ -5622,7 +5621,7 @@
   ARMTargetCodeGenInfo::setTargetAttributes(D, GV, CGM, IsForDefinition);
   if (!IsForDefinition)
 return;
-  addStackProbeSizeTargetAttribute(D, GV, CGM);
+  addStackProbeTargetAttributes(D, GV, CGM);
 }
 }
 
Index: include/clang/Frontend/CodeGenOptions.def
===
--- include/clang/Frontend/CodeGenOptions.def
+++ include/clang/Frontend/CodeGenOptions.def
@@ -223,6 +223,7 @@
 ///< alignment, if not 0.
 VALUE_CODEGENOPT(StackProbeSize, 32, 4096) ///< Overrides default stack
///< probe size, even if 0.
+CODEGENOPT(NoStackArgProbe, 1, 0) ///< Set when -mno-stack-arg-probe is used
 CODEGENOP

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added inline comments.



Comment at: lib/Driver/ToolChains/Clang.cpp:4038
+  if (Args.hasArg(options::OPT_mno_stack_arg_probe))
+CmdArgs.push_back("-mno-stack-arg-probe");
+

hans wrote:
> I think you can just do
> 
> ```
> Args.AddLastArg(CmdArgs, options::OPT_mno_stack_arg_probe)
> ```
> 
> which avoids the if-statement and having to spell out the flag.
I added mstack_arg_probe also, and this check now contains two options. I am 
not sure how AddLastArg behaves in this case (especially because only one 
attribute needs to be defined: mno-stack-arg-probe, and a positive case is the 
default). I found some other examples similar to this one, and they used hasArg 
with two options.


https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@hans: I responded to the comments and also updated diff for clang with new 
changes.


https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@hans, @craig.topper, @MatzeB: Can someone commit 
https://reviews.llvm.org/D43108 (this) and https://reviews.llvm.org/D43107 on 
my behalf? I do not think, I have commit access.


https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@aemerson : I am not 100% sure, but I think you are talking about different 
flag. This commit is for supporting a flag to disable stack probes (identical 
flag can be found in MinGW), it basically only applies to PE/COFF (Windows) 
targets. It can be useful to compile UEFI code. By default, Windows will use 
check probes of 4K in x86. A related stack-probe-size allows to vary the size 
of stack probes. This one completely disables them.

By default, stack probes are enabled (i.e., -mstack-arg-probe is the default 
behavior) and have the size of 4K in x86.


Repository:
  rC Clang

https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D43108: Support for the mno-stack-arg-probe flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@aemerson : yes, it is just part of MS ABI


Repository:
  rC Clang

https://reviews.llvm.org/D43108



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan created this revision.
nruslan added reviewers: hans, craig.topper.
Herald added a subscriber: cfe-commits.

Allows to disable direct TLS segment access (%fs or %gs). GCC supports a 
similar flag, it can be useful in some circumstances, e.g. when a thread 
context block needs to be updated directly from user space. More info and 
specific use cases: https://bugs.llvm.org/show_bug.cgi?id=16145

There is another revision for LLVM as well.


Repository:
  rC Clang

https://reviews.llvm.org/D53102

Files:
  docs/ClangCommandLineReference.rst
  include/clang/Driver/CC1Options.td
  include/clang/Driver/Options.td
  include/clang/Frontend/CodeGenOptions.def
  lib/CodeGen/CGCall.cpp
  lib/Driver/ToolChains/Clang.cpp
  lib/Frontend/CompilerInvocation.cpp
  test/CodeGen/indirect-tls-seg-refs.c
  test/Driver/indirect-tls-seg-refs.c

Index: test/Driver/indirect-tls-seg-refs.c
===
--- /dev/null
+++ test/Driver/indirect-tls-seg-refs.c
@@ -0,0 +1,7 @@
+// RUN: %clang -### %s 2>&1 | FileCheck %s -check-prefix=TLSDIRECT
+// RUN: %clang -### -mno-tls-direct-seg-refs -mtls-direct-seg-refs %s 2>&1 | FileCheck %s -check-prefix=TLSDIRECT
+// RUN: %clang -### -mtls-direct-seg-refs -mno-tls-direct-seg-refs %s 2>&1 | FileCheck %s -check-prefix=NO-TLSDIRECT
+// REQUIRES: clang-driver
+
+// NO-TLSDIRECT: -indirect-tls-seg-refs
+// TLSDIRECT-NOT: -indirect-tls-seg-refs
Index: test/CodeGen/indirect-tls-seg-refs.c
===
--- /dev/null
+++ test/CodeGen/indirect-tls-seg-refs.c
@@ -0,0 +1,8 @@
+// RUN: %clang_cc1 %s -emit-llvm -o - -indirect-tls-seg-refs | FileCheck %s -check-prefix=NO-TLSDIRECT
+// RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s -check-prefix=TLSDIRECT
+
+// NO-TLSDIRECT: attributes #{{[0-9]+}} = {{{.*}} "indirect-tls-seg-refs"
+// TLSDIRECT-NOT: attributes #{{[0-9]+}} = {{{.*}} "indirect-tls-seg-refs"
+
+void test1() {
+}
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -614,6 +614,7 @@
   Opts.DisableLifetimeMarkers = Args.hasArg(OPT_disable_lifetimemarkers);
   Opts.DisableO0ImplyOptNone = Args.hasArg(OPT_disable_O0_optnone);
   Opts.DisableRedZone = Args.hasArg(OPT_disable_red_zone);
+  Opts.IndirectTlsSegRefs = Args.hasArg(OPT_indirect_tls_seg_refs);
   Opts.ForbidGuardVariables = Args.hasArg(OPT_fforbid_guard_variables);
   Opts.UseRegisterSizedBitfieldAccess = Args.hasArg(
 OPT_fuse_register_sized_bitfield_access);
Index: lib/Driver/ToolChains/Clang.cpp
===
--- lib/Driver/ToolChains/Clang.cpp
+++ lib/Driver/ToolChains/Clang.cpp
@@ -1739,6 +1739,10 @@
   Args.hasArg(options::OPT_fapple_kext))
 CmdArgs.push_back("-disable-red-zone");
 
+  if (!Args.hasFlag(options::OPT_mtls_direct_seg_refs,
+  options::OPT_mno_tls_direct_seg_refs, true))
+CmdArgs.push_back("-indirect-tls-seg-refs");
+
   // Default to avoid implicit floating-point for kernel/kext code, but allow
   // that to be overridden with -mno-soft-float.
   bool NoImplicitFloat = (Args.hasArg(options::OPT_mkernel) ||
Index: lib/CodeGen/CGCall.cpp
===
--- lib/CodeGen/CGCall.cpp
+++ lib/CodeGen/CGCall.cpp
@@ -1709,6 +1709,8 @@
 
   if (CodeGenOpts.DisableRedZone)
 FuncAttrs.addAttribute(llvm::Attribute::NoRedZone);
+  if (CodeGenOpts.IndirectTlsSegRefs)
+FuncAttrs.addAttribute("indirect-tls-seg-refs");
   if (CodeGenOpts.NoImplicitFloat)
 FuncAttrs.addAttribute(llvm::Attribute::NoImplicitFloat);
 
Index: include/clang/Frontend/CodeGenOptions.def
===
--- include/clang/Frontend/CodeGenOptions.def
+++ include/clang/Frontend/CodeGenOptions.def
@@ -62,6 +62,8 @@
 CODEGENOPT(DebugPassManager, 1, 0) ///< Prints debug information for the new
///< pass manager.
 CODEGENOPT(DisableRedZone, 1, 0) ///< Set when -mno-red-zone is enabled.
+CODEGENOPT(IndirectTlsSegRefs, 1, 0) ///< Set when -mno-tls-direct-seg-refs
+ ///< is specified.
 CODEGENOPT(DisableTailCalls  , 1, 0) ///< Do not emit tail calls.
 CODEGENOPT(NoEscapingBlockTailCalls, 1, 0) ///< Do not emit tail calls from
///< escaping blocks.
Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -2006,6 +2006,8 @@
 def mno_pascal_strings : Flag<["-"], "mno-pascal-strings">,
   Alias;
 def mno_red_zone : Flag<["-"], "mno-red-zone">, Group;
+def mno_tls_direct_seg_refs : Flag<["-"], "mno-tls-direct-seg-refs">, Group,
+  HelpText<"Disable direct TLS access through segment registers">;

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added inline comments.



Comment at: docs/ClangCommandLineReference.rst:2241
+
+Enable or disable direct TLS access through segment registers
+

hans wrote:
> This file is automatically generated based on the options .td files, so no 
> need to update it here.
ok



Comment at: include/clang/Driver/CC1Options.td:195
   HelpText<"Do not emit code that uses the red zone.">;
+def indirect_tls_seg_refs : Flag<["-"], "indirect-tls-seg-refs">,
+  HelpText<"Do not emit code that uses direct TLS segment access.">;

hans wrote:
> Could mno_tls_direct_seg_refs be used as a cc1 flag instead?
done



Comment at: include/clang/Driver/Options.td:2167
+def mtls_direct_seg_refs : Flag<["-"], "mtls-direct-seg-refs">, Group,
+  HelpText<"Enable direct TLS access through segment registers">;
 def mregparm_EQ : Joined<["-"], "mregparm=">, Group;

hans wrote:
> Maybe add (default) to the help text to indicate this is the default?
done


https://reviews.llvm.org/D53102



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan updated this revision to Diff 169224.
nruslan marked 3 inline comments as done.
nruslan added a comment.

@hans , @craig.topper : Updated the patch with all requested changes


https://reviews.llvm.org/D53102

Files:
  include/clang/Driver/Options.td
  include/clang/Frontend/CodeGenOptions.def
  lib/CodeGen/CGCall.cpp
  lib/Driver/ToolChains/Clang.cpp
  lib/Frontend/CompilerInvocation.cpp
  test/CodeGen/indirect-tls-seg-refs.c
  test/Driver/indirect-tls-seg-refs.c


Index: test/Driver/indirect-tls-seg-refs.c
===
--- /dev/null
+++ test/Driver/indirect-tls-seg-refs.c
@@ -0,0 +1,7 @@
+// RUN: %clang -### %s 2>&1 | FileCheck %s -check-prefix=TLSDIRECT
+// RUN: %clang -### -mno-tls-direct-seg-refs -mtls-direct-seg-refs %s 2>&1 | 
FileCheck %s -check-prefix=TLSDIRECT
+// RUN: %clang -### -mtls-direct-seg-refs -mno-tls-direct-seg-refs %s 2>&1 | 
FileCheck %s -check-prefix=NO-TLSDIRECT
+// REQUIRES: clang-driver
+
+// NO-TLSDIRECT: -mno-tls-direct-seg-refs
+// TLSDIRECT-NOT: -mno-tls-direct-seg-refs
Index: test/CodeGen/indirect-tls-seg-refs.c
===
--- /dev/null
+++ test/CodeGen/indirect-tls-seg-refs.c
@@ -0,0 +1,8 @@
+// RUN: %clang_cc1 %s -emit-llvm -o - -mno-tls-direct-seg-refs | FileCheck %s 
-check-prefix=NO-TLSDIRECT
+// RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s -check-prefix=TLSDIRECT
+
+// NO-TLSDIRECT: attributes #{{[0-9]+}} = {{{.*}} "indirect-tls-seg-refs"
+// TLSDIRECT-NOT: attributes #{{[0-9]+}} = {{{.*}} "indirect-tls-seg-refs"
+
+void test1() {
+}
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -614,6 +614,7 @@
   Opts.DisableLifetimeMarkers = Args.hasArg(OPT_disable_lifetimemarkers);
   Opts.DisableO0ImplyOptNone = Args.hasArg(OPT_disable_O0_optnone);
   Opts.DisableRedZone = Args.hasArg(OPT_disable_red_zone);
+  Opts.IndirectTlsSegRefs = Args.hasArg(OPT_mno_tls_direct_seg_refs);
   Opts.ForbidGuardVariables = Args.hasArg(OPT_fforbid_guard_variables);
   Opts.UseRegisterSizedBitfieldAccess = Args.hasArg(
 OPT_fuse_register_sized_bitfield_access);
Index: lib/Driver/ToolChains/Clang.cpp
===
--- lib/Driver/ToolChains/Clang.cpp
+++ lib/Driver/ToolChains/Clang.cpp
@@ -1739,6 +1739,10 @@
   Args.hasArg(options::OPT_fapple_kext))
 CmdArgs.push_back("-disable-red-zone");
 
+  if (!Args.hasFlag(options::OPT_mtls_direct_seg_refs,
+  options::OPT_mno_tls_direct_seg_refs, true))
+CmdArgs.push_back("-mno-tls-direct-seg-refs");
+
   // Default to avoid implicit floating-point for kernel/kext code, but allow
   // that to be overridden with -mno-soft-float.
   bool NoImplicitFloat = (Args.hasArg(options::OPT_mkernel) ||
Index: lib/CodeGen/CGCall.cpp
===
--- lib/CodeGen/CGCall.cpp
+++ lib/CodeGen/CGCall.cpp
@@ -1709,6 +1709,8 @@
 
   if (CodeGenOpts.DisableRedZone)
 FuncAttrs.addAttribute(llvm::Attribute::NoRedZone);
+  if (CodeGenOpts.IndirectTlsSegRefs)
+FuncAttrs.addAttribute("indirect-tls-seg-refs");
   if (CodeGenOpts.NoImplicitFloat)
 FuncAttrs.addAttribute(llvm::Attribute::NoImplicitFloat);
 
Index: include/clang/Frontend/CodeGenOptions.def
===
--- include/clang/Frontend/CodeGenOptions.def
+++ include/clang/Frontend/CodeGenOptions.def
@@ -62,6 +62,8 @@
 CODEGENOPT(DebugPassManager, 1, 0) ///< Prints debug information for the new
///< pass manager.
 CODEGENOPT(DisableRedZone, 1, 0) ///< Set when -mno-red-zone is enabled.
+CODEGENOPT(IndirectTlsSegRefs, 1, 0) ///< Set when -mno-tls-direct-seg-refs
+ ///< is specified.
 CODEGENOPT(DisableTailCalls  , 1, 0) ///< Do not emit tail calls.
 CODEGENOPT(NoEscapingBlockTailCalls, 1, 0) ///< Do not emit tail calls from
///< escaping blocks.
Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -2006,6 +2006,8 @@
 def mno_pascal_strings : Flag<["-"], "mno-pascal-strings">,
   Alias;
 def mno_red_zone : Flag<["-"], "mno-red-zone">, Group;
+def mno_tls_direct_seg_refs : Flag<["-"], "mno-tls-direct-seg-refs">, 
Group, Flags<[CC1Option]>,
+  HelpText<"Disable direct TLS access through segment registers">;
 def mno_relax_all : Flag<["-"], "mno-relax-all">, Group;
 def mno_rtd: Flag<["-"], "mno-rtd">, Group;
 def mno_soft_float : Flag<["-"], "mno-soft-float">, Group;
@@ -2161,6 +2163,8 @@
 def moslib_EQ : Joined<["-"], "moslib=">, Group;
 def mpascal_strings : Flag<["-"], "mpascal-strings">, Alias;
 def mred_zone : Flag<["-"], "mr

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@hans, @craig.topper : If everything is fine, can someone check in the clang 
and llvm patches on my behalf? I do not have commit access.


https://reviews.llvm.org/D53102



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

In https://reviews.llvm.org/D53102#1268272, @kristina wrote:

> If the author doesn't mind I can just apply the style fix after patching and 
> then rebuild and run all the relevant tests (or would you prefer to do 
> that?). Seems easier than a new revision for an indentation change on one 
> line.


@kristina : Thank you! Sure, that is fine. Go ahead.


https://reviews.llvm.org/D53102



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

In https://reviews.llvm.org/D53102#1268364, @kristina wrote:

> By the way, out of curiosity is this for anything specific (alternative libc 
> or some user-mode-scheduling implementation)? Not nitpicking, just curious 
> since it's an interesting topic in general and it's frustrating that the 
> architecture is so limited in terms of registers that can be used for 
> TLS/related data unlike newer ARM/ARM64 architectures.
>
> Also FWIW `%gs` is generally free to use under x86_64 Linux which is where I 
> usually place my thread control blocks which doesn't interfere with libc 
> which uses `%fs`. (Just started build/test job, waiting on that for now)


@kristina : Yes, there are some recent use cases mentioned in the bug report 
(see the link above). Also, I know that it was used for x86 (32-bit only) Xen 
guests where it would be recommended to compile an OS with this flag. I also 
used it in my fork of the NPTL pthread library in VirtuOS to support M:N 
threading model: http://sigops.org/sosp/sosp13/papers/p116-nikolaev.pdf


https://reviews.llvm.org/D53102



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53102: Support for the mno-tls-direct-seg-refs flag

[Ruslan Nikolaev via Phabricator via cfe-commits]

nruslan added a comment.

@kristina : Thank you very much for taking care of the patchsets!


Repository:
  rC Clang

https://reviews.llvm.org/D53102



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits