Opened 3 weeks ago

Closed 3 weeks ago

Last modified 3 weeks ago

#70790 closed defect (invalid)

m4 build error on Sequoia

Reported by: eschnett (Erik Schnetter) Owned by:
Priority: Normal Milestone:
Component: ports Version: 2.10.1
Keywords: Sequoia Cc:
Port: m4

Description (last modified by jmroot (Joshua Root))

Building m4 on Sequoia fails in the configure stage. The relevant lines are

| #include <sys/types.h>
|                    #include <sys/socket.h>
| 
|                    int getpeername (int, void *, unsigned long int *);
| int
| main (void)
| {
| unsigned long int len;
|                   getpeername (0, 0, &len);
|   ;
|   return 0;
| }
configure:58033: error: Cannot find a type to use in place of socklen_t

I believe the reason is the following. The compiler is invoked with

configure:58023: /usr/bin/clang -c -pipe -Os -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk -arch arm64 -I/opt/local/include -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk conftest.c >&5

but the file sys/socket.h is not located in this sysroot. Instead, it is found at /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include/sys/socket.h.

Attachments (1)

config.log.bz2 (88.6 KB) - added by jmroot (Joshua Root) 3 weeks ago.

Download all attachments as: .zip

Change History (6)

Changed 3 weeks ago by jmroot (Joshua Root)

Attachment: config.log.bz2 added

comment:1 Changed 3 weeks ago by jmroot (Joshua Root)

Description: modified (diff)

Something seems to be wrong with the SDK you have. On my system:

% ls -l /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/sys/socket.h 
-r--r--r--  1 root  wheel  32534 13 Aug 14:03 /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/sys/socket.h

Your machine/endian.h also seems to be including i386/endian.h when building for arm:

configure:17916: /usr/bin/clang -o conftest -pipe -Os -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk -arch arm64 -I/opt/local/include -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk -L/opt/local/lib -Wl,-headerpad_max_install_names -Wl,-syslibroot,/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk -arch arm64 conftest.c  >&5
In file included from conftest.c:155:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/stdlib.h:58:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/_stdlib.h:66:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/sys/wait.h:186:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/machine/endian.h:35:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/i386/endian.h:85:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/sys/_endian.h:131:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/libkern/_OSByteOrder.h:62:
/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/libkern/arm/_OSByteOrder.h:50:1: error: redefinition of '_OSSwapInt16'
   50 | _OSSwapInt16(
      | ^
/Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk/usr/include/libkern/i386/_OSByteOrder.h:50:1: note: previous definition is here
   50 | _OSSwapInt16(
      | ^

comment:2 Changed 3 weeks ago by eschnett (Erik Schnetter)

Indeed I see on my system

uname -m = x86_64

which is wrong (I have an M3 chip). Investigating...

comment:3 Changed 3 weeks ago by eschnett (Erik Schnetter)

It turns out that my terminal emulator, and possibly also my shell, were x86_64 executables running with Rosetta. Maybe this confuses the install scripts? I thought that the default system architecture would still be arm64, but apparently something was off.

Anyway, I'm now using a native terminal emulator and a native shell. All is working fine, and I was able to install m4 without problems.

Thanks for spotting the i386 clue.

comment:4 Changed 3 weeks ago by cjones051073 (Chris Jones)

Resolution: invalid
Status: newclosed

comment:5 Changed 3 weeks ago by jmroot (Joshua Root)

That's still quite strange, since the -arch arm64 option should make clang build for arm64 regardless of what architecture the program itself is running as. I guess you have your fix, and running natively is certainly preferable anyway, but please feel free to reopen if anyone figures out more about this and knows how we can prevent such problems when running via Rosetta.

Note: See TracTickets for help on using tickets.