From: herbert@laubners.info Subject: Re: darktable 4.8.1 To: Stuart Henderson Cc: "ports@openbsd.org" Date: Fri, 15 Aug 2025 17:52:32 +0800 > owner-ports@openbsd.org hat am 15.08.2025 16:51 +08 geschrieben: > > > Your post to ports@openbsd.org must be confirmed because > herbert@laubners.info > is not a member of the mailing list. > > To accept the posting, please do one of the following: > > 1. If you have web browsing capability, visit > > > 2. Reply to majordomo@openbsd.org > with the following command in the body of the message: > > accept > > (The number 8235-1A7E-D7AF must be in the Subject header) > > 3. Reply to majordomo@openbsd.org > with the following command in the body of the message: > > accept 8235-1A7E-D7AF > > If you do not respond within 7 days, this token will expire, > and the request will not be completed. > > If you are already a member of the ports mailing list, you > may add the address you are posting from as an "alias" via the web > interface at https://lists.openbsd.org/cgi-bin/mj_wwwusr?func=show > > If you would like to communicate with a person, > send mail to owner-ports@openbsd.org. > Stuart Henderson hat am 15.08.2025 15:29 +08 geschrieben: > > > On 2025/08/15 10:26, herbert@laubners.info wrote: > > OS: OpenBSD 7.7 amd64 > > Host: ASUS TUF GAMING B550-PLUS > > CPU: AMD Ryzen 5 5600G with Radeon Graphics (12) @ 3.900GHz > > GPU: AMD Radeon Graphics (radeonsi, renoir, LLVM 16.0.6, DRM 3.61, 7.7) > > Memory: 1069MiB / 48378MiB > > > > Darktable crashes with segmentation fault, indipendend if installed from packages or ports. > > Option opencl activated during start does not change it. > > > .. > > > Is there anything else, I can provide? > > steps to reproduce? I had the same same setup (files on NAS in local network) earlier on Gentoo, darktable was on version 5.2.x I am thinking on teh calculations done the rdd.c file, since the segmentation fault stops there: Thread 6 received signal SIGSEGV, Segmentation fault. [Switching to thread 478363 of process 16229] 0x00000ad90d910d52 in rcd_ppg_border (out=0x0, in=0xad99ebbe000, width=, height=2865, filters=2492765332, margin=7) at /usr/ports/pobj/darktable-4.8.1/darktable-4.8.1/src/iop/demosaicing/rcd.c:122 122 out[4 * ((size_t)j * width + i) + c] = fmaxf(0.0f, in[(size_t)j * width + i]); Is it posisble that cc is using another size of 'const int with' then gcc under linux and with that the index for the scalar out[] is outside of assigned menory?