On Wed Mar 18 16:34:37 2026 +0000, Giovanni Mascellani wrote:
Thanks for investigating this. Would you consider this as a fixup? https://gitlab.winehq.org/giomasce/wine/-/commit/0c342aeb3f61c7455487deb684c... Basically it integrates the check with the code immediately following, hopefully making it more linear to understand. It also does the same change to other similar or identical copies of that function. Hello, thanks for taking a look. I put your fixup patch on top of the ASan tree, but received now this failure: https://gitlab.winehq.org/bernhardu/wine/-/jobs/239086#L2792
To make handling this easier, is it ok to squash it into one patch, and add a "Co-authored-by: Giovanni Mascellani <gmascellani@codeweavers.com>"? Is this probably an issue of `WINMM_OpenDevice` by handling all `info->format->wFormatTag` different than `WAVE_FORMAT_PCM` with a `malloc(sizeof(WAVEFORMATEX) + info->format->cbSize)`? Should there be a check for `info->format->wFormatTag == WAVE_FORMAT_EXTENSIBLE`, which then allocates at least the size of `WAVEFORMATEXTENSIBLE`? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10326#note_132653