[lldb][AIX] Adding AIX version of ptrace64

[DhruvSrivastavaX]

This PR is in reference to porting LLDB on AIX.

Link to discussions on llvm discourse and github:

1. https://discourse.llvm.org/t/port-lldb-to-ibm-aix/80640
2. Extending LLDB to work on AIX #101657
   The complete changes for porting are present in this draft PR:
   Extending LLDB to work on AIX #102601

Adding changes for minimal build for lldb binary on AIX. ptrace64 is needed to debug 64-bit AIX debuggee, and its format is different than the traditional ptrace on other platforms: AIX ptrace

Review Request: @labath @DavidSpickett

[llvmbot]

@llvm/pr-subscribers-lldb

Author: Dhruv Srivastava (DhruvSrivastavaX)

Changes

This PR is in reference to porting LLDB on AIX.

Link to discussions on llvm discourse and github:

1. https://discourse.llvm.org/t/port-lldb-to-ibm-aix/80640
2. Extending LLDB to work on AIX #101657
   The complete changes for porting are present in this draft PR:
   Extending LLDB to work on AIX #102601

Adding changes for minimal build for lldb binary on AIX.
ptrace64 is needed to debug 64-bit AIX debuggee, and its format is different than the traditional ptrace on other platforms:
https://www.ibm.com/docs/fr/aix/7.3?topic=p-ptrace-ptracex-ptrace64-subroutine

Review Request: @labath @DavidSpickett

---

Full diff: https://github.com/llvm/llvm-project/pull/120390.diff

1 Files Affected:

• (modified) lldb/source/Host/posix/ProcessLauncherPosixFork.cpp (+7-3)

diff --git a/lldb/source/Host/posix/ProcessLauncherPosixFork.cpp b/lldb/source/Host/posix/ProcessLauncherPosixFork.cpp
index 4a9469bde2f186..0e162d04c35837 100644
--- a/lldb/source/Host/posix/ProcessLauncherPosixFork.cpp
+++ b/lldb/source/Host/posix/ProcessLauncherPosixFork.cpp
@@ -21,8 +21,8 @@
 #include <sys/wait.h>
 #include <unistd.h>
 
-#include <sstream>
 #include <csignal>
+#include <sstream>
 
 #ifdef __ANDROID__
 #include <android/api-level.h>
@@ -47,8 +47,7 @@ static void write_string(int error_fd, const char *str) {
   (void)r;
 }
 
-[[noreturn]] static void ExitWithError(int error_fd,
-                                       const char *operation) {
+[[noreturn]] static void ExitWithError(int error_fd, const char *operation) {
   int err = errno;
   write_string(error_fd, operation);
   write_string(error_fd, " failed: ");
@@ -193,8 +192,13 @@ struct ForkLaunchInfo {
     }
 
     // Start tracing this child that is about to exec.
+#if !defined(_AIX)
     if (ptrace(PT_TRACE_ME, 0, nullptr, 0) == -1)
       ExitWithError(error_fd, "ptrace");
+#else
+    if (ptrace64(PT_TRACE_ME, 0, 0, 0, nullptr) == -1)
+      ExitWithError(error_fd, "ptrace");
+#endif
   }
 
   // Execute.  We should never return...

[DavidSpickett]

Please link to the English docs - https://www.ibm.com/docs/en/aix/7.3?topic=p-ptrace-ptracex-ptrace64-subroutine instead.

[DhruvSrivastavaX]

Please link to the English docs - https://www.ibm.com/docs/en/aix/7.3?topic=p-ptrace-ptracex-ptrace64-subroutine instead.

Apologies for the blunder, I myself dont speak French by the way. 😀 Corrected

[DavidSpickett]

Can you summarise the need for ptrace64? Mainly, will this only apply to launching a process or does it apply to any ptrace call?

I wonder:

1. How many #ifdef we're going to end up with
2. Whether AIX should use ptrace64 everywhere if your intention is to only support AIX 64 bit (but if it's only this one use of ptrace that needs ptrace64, that's not necessary)

[DavidSpickett]

Apologies for the blunder, I myself dont speak French by the way. 😀 Corrected

I am kinda impressed how many languages they have available, but the downside is that Google seems to love choosing a non-English one for me.

[DhruvSrivastavaX]

I am kinda impressed how many languages they have available, but the downside is that Google seems to love choosing a non-English one for me.

True. Pretty much same for me. It will always be either Japanese or Korean. French is a new addition perhaps. 😄

[DhruvSrivastavaX]

Can you summarise the need for ptrace64? Mainly, will this only apply to launching a process or does it apply to any ptrace call?

I wonder:

1. How many #ifdef we're going to end up with
2. Whether AIX should use ptrace64 everywhere if your intention is to only support AIX 64 bit (but if it's only this one use of ptrace that needs ptrace64, that's not necessary)

Yes, to handle the 64-bit debuggees in AIX, we would need the ptrace64 call. It will apply to all the ptrace calls in general.
It is used in the NativeProcess/Thread plugins, but since those are new plugins only for AIX they dont need an #if AIX.
One other place that will need an #if AIX besides the one in this PR is GDBRemoteCommunicationServerLLGS.cpp and that is pretty much it for now.

[DhruvSrivastavaX]

AIX ptrace (ptrace,ptracex,ptrace64) has a different format, so we would need an AIX specific case anyway.

[DavidSpickett]

Yes, to handle the 64-bit debuggees in AIX, we would need the ptrace64 call. It will apply to all the ptrace calls in general.
It is used in the NativeProcess/Thread plugins, but since those are new plugins only for AIX they dont need an #if AIX.

Thanks. Yes I see that the majority of existing calls are in, for example, lldb/source/Plugins/Process/Linux/NativeProcessLinux.cpp. And a lot are going via a Ptracewrapper that is also per platform.

This PosixFork is one of the few doing a direct call in generic code.

I am wondering if it should be #if there is ptrace64 but I don't think so because this seems to be AIX specific. Even if it existed elsewhere I'd rather not silently use it without knowing how it behaves.

[DavidSpickett]

@labath what do you think?

[labath]

I think this is fine. We could come up with an abstraction for this, but I don't think it's necessary for a simple change like this. Like @DhruvSrivastavaX said, the rest of the ptrace calls are going to be in platform-specific code.