RFR: JDK-8354316 : clang/linux build fails with -Wunused-result warning at XToolkit.c:695:9

Sergey Bylokhov serb at openjdk.org
Wed May 14 04:25:52 UTC 2025


On Tue, 13 May 2025 18:36:18 GMT, Alexey Ivanov <aivanov at openjdk.org> wrote:

>> The following line results in unused-result warning on linux/clang.
>> 
>> 
>> /java.desktop/unix/native/libawt_xawt/xawt/XToolkit.c:695:9: error: ignoring return value of function
>>  declared with 'warn_unused_result' attribute [-Werror,-Wunused-result]
>>  695 | write ( AWT_WRITEPIPE, &wakeUp_char, 1 ); 
>> 
>> 
>> There are two ways to handle it 
>> 
>> 1) Make changes to XToolkit.c such that the warning is no longer thrown. But throwing an error based on the result of `write ( AWT_WRITEPIPE, &wakeUp_char, 1 );` will result in unexpected behavioral changes and the best way to handle it is to have an empty if block with an appropriate comment.
>> 
>> 
>> There was a discussion about the same line long ago and the reason the result of `write()` was not handled and it was left unchanged was not to introduce behavioral change - https://mail.openjdk.org/pipermail/awt-dev/2016-July/011626.html
>> 
>> 
>> 2) Add unused-result to disabled warning section for clang similar to gcc - https://github.com/openjdk/jdk/blob/d1543429ff29ca0d761b8473b3fb8621abcd226d/make/modules/java.desktop/lib/AwtLibraries.gmk#L281. The 1st approach was picked over the 2nd since the usual recommendation is not to add to disabled warning section unless there is no other option.
>> 
>> NOTE: the fix has been tested on linux/gcc , it does need to be tested on linux/clang.
>
> src/java.desktop/unix/native/libawt_xawt/xawt/XToolkit.c line 696:
> 
>> 694:     if (!isMainThread() && awt_pipe_inited) {
>> 695:         if (write(AWT_WRITEPIPE, &wakeUp_char, 1) < 0) {
>> 696:             // if block is left empty to avoid adding unused-result to
> 
> Does it make sense to add a trace with `DTRACE_PRINTLN*`?
> 
> It could be useful for debugging…

yes If write does not return 1 might indicate an error, it would be good to log it. so we can skip this long comment

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/25217#discussion_r2088003281


More information about the client-libs-dev mailing list