Bug 114970 - [14/15 Regression] 32-bit ARM gcc-14.1 new false positive -Wunused-value
Summary: [14/15 Regression] 32-bit ARM gcc-14.1 new false positive -Wunused-value
Status: UNCONFIRMED
Alias: None
Product: gcc
Classification: Unclassified
Component: c++ (show other bugs)
Version: 14.1.0
: P3 normal
Target Milestone: 14.2
Assignee: Not yet assigned to anyone
URL:
Keywords: diagnostic
Depends on:
Blocks:
 
Reported: 2024-05-07 10:51 UTC by Nils Gladitz
Modified: 2024-05-07 21:33 UTC (History)
2 users (show)

See Also:
Host:
Target: arm*-*-*eabi
Build:
Known to work:
Known to fail:
Last reconfirmed:

Attachments
Testcase (141 bytes, text/x-csrc)
2024-05-07 10:51 UTC, Nils Gladitz 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Nils Gladitz 2024-05-07 10:51:13 UTC 
Created attachment 58114 [details]
Testcase

The attached reduced test case produces the following warning:
test.cpp:6:43: warning: value computed is not used [-Wunused-value]
    6 |         std::string_view prefix("<prefix>");

When compiling with:
 -Wall -std=c++20 -O2

I see the warning when building for 32-bit ARM but not x86-64.
There is no warning with GCC 13.2.0.

Also seen with godbolt on "trunk": https://godbolt.org/z/GvbMMsfch
Comment 1 Andrew Pinski 2024-05-07 20:44:06 UTC 
This looks related to constructor return this where with ARM EABI constructors (and deconstructors return this) and we produce:
      <<cleanup_point <<< Unknown tree: expr_stmt
        prefix = {._M_len=8, ._M_str=(const char *) "<prefix>"}, (struct basic_string_view *) &prefix >>>>>;


Note "this" of &prefix is there.

Let me try to reduce this a little bit.
I might not get to it until week after next though.