[28.x] build: suppress -Wunterminated-string-initialization in secp256k1 #32484

1. 
   fanquake commented at 4:18 pm on May 13, 2025: member

   Suppress -Wunterminated-string-initialization warnings from GCC 15, like:

    0In file included from src/secp256k1.c:830:
    1src/modules/schnorrsig/main_impl.h:48:46: warning: initializer-string for array of 'unsigned char' truncates NUL terminator but destination lacks 'nonstring' attribute (14 chars into 13 available) [-Wunterminated-string-initialization]
    2   48 | static const unsigned char bip340_algo[13] = "BIP0340/nonce";
    3      |                                              ^~~~~~~~~~~~~~~
    4In file included from src/tests_exhaustive.c:28:
    5src/testrand_impl.h: In function ‘testrand_seed’:
    6src/testrand_impl.h:21:45: warning: initializer-string for array of ‘unsigned char’ truncates NUL terminator but destination lacks ‘nonstring’ attribute (20 chars into 19 available) [-Wunterminated-string-initialization]
    7   21 |     static const unsigned char PREFIX[19] = "secp256k1 test init";
    8      |                                             ^~~~~~~~~~~~~~~~~~~~~
    9src/modules/ellswift/tests_impl.h:411:53: warning: initializer-string for array of ‘unsigned char’ truncates NUL terminator but destination lacks ‘nonstring’ attribute (27 chars into 26 available) [-Wunterminated-string-initialization]
   10  411 |         static const unsigned char bip324_tag[26] = "bip324_ellswift_xonly_ecdh";
   11      |                                                     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
   

   This was fixed upstream in https://github.com/bitcoin-core/secp256k1/pull/1583, so only relevant for 28.x and later. The code is fine, but the warnings may break builds under -Werror, leave users wondering if there is a real issue etc.

2. 
   DrahtBot commented at 4:18 pm on May 13, 2025: contributor

   The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

   Code Coverage & Benchmarks

   For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/32484.

   Reviews

   See the guideline for information on the review process.

   Type	Reviewers
   Stale ACK	hebasto

   If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.

3. DrahtBot added the label Backport on May 13, 2025
4. 
   theuni commented at 4:49 pm on May 13, 2025: member

   Hmm, my first thought was just to backport the “fix” instead, but I guess we’ve never really backported secp changes for the sake of keeping in sync.

   Either way, in this case I agree it’s simple/safe enough to just suppress the warning for 28.x.

5. hebasto approved
6. 
   hebasto commented at 4:53 pm on May 13, 2025: member

   ACK 5c8a55f917d089c8bdf5f59195116d9201c597e7, tested on Ubuntu 25.04 with GCC-15.
7. fanquake added this to the milestone 28.2 on May 13, 2025
8. DrahtBot added the label CI failed on May 13, 2025
9. [28.x] build: suppress -Wunterminated-string-initialization in secp

   Suppress `-Wunterminated-string-initialization` warnings from GCC 15,
   like:
   ```bash
   In file included from src/secp256k1.c:830:
   src/modules/schnorrsig/main_impl.h:48:46: warning: initializer-string for array of 'unsigned char' truncates NUL terminator but destination lacks 'nonstring' attribute (14 chars into 13 available) [-Wunterminated-string-initialization]
      48 | static const unsigned char bip340_algo[13] = "BIP0340/nonce";
         |                                              ^~~~~~~~~~~~~~~
   In file included from src/tests_exhaustive.c:28:
   src/testrand_impl.h: In function ‘testrand_seed’:
   src/testrand_impl.h:21:45: warning: initializer-string for array of ‘unsigned char’ truncates NUL terminator but destination lacks ‘nonstring’ attribute (20 chars into 19 available) [-Wunterminated-string-initialization]
      21 |     static const unsigned char PREFIX[19] = "secp256k1 test init";
         |                                             ^~~~~~~~~~~~~~~~~~~~~
   ```
   
   This was fixed upstream in
   https://github.com/bitcoin-core/secp256k1/pull/1583.

   7fc15db5f9
10. fanquake force-pushed on May 16, 2025
11. 
    fanquake commented at 2:12 pm on May 16, 2025: member

    Changed this, as the prior approach would warn under Clang.
12. 
    maflcko commented at 3:35 pm on May 16, 2025: member

    An alternative would be to just ignore the warning? CI on 28.x isn’t using gcc 15 and users aren’t really expected to compile with werror?
13. 
    fanquake commented at 3:42 pm on May 16, 2025: member

    We could also do that. I guess it depends where we expect this branch to be compiled. Any downstream CIs, build systems, packagers etc may be turning warnings into errors. It doesn’t seem bad for us to suppress known non-issues.

Contributors
fanquake DrahtBot theuni hebasto maflcko

Labels
Backport CI failed

Milestone
28.2