Hi, I'm on rawhide with latest updates, except for some that have package version conflicts. But, the latest gcc and glibc packages. First, when I was configuring, (make menuconfig), even though it said that sysfb and sysfb-simplefb were in /drivers/firmware, they weren't there. They took the setting from the last 5.16 kernel config that I used as a template. Then, when I tried to compile the 5.17 kernel from kernel-5.17.0-0.rc2.20220204gitdcb85f85fa6f.86.fc36.src.rpm it failed because I had warnings as errors, and the kernel warned about ssh version being deprecated for openssl 3. e.g. scripts/sign-file.c: In function 'read_private_key': scripts/sign-file.c:142:17: warning: 'ENGINE_load_builtin_engines' is deprecated: Since OpenSSL 3.0 [-Wdeprecated-declarations] 142 | ENGINE_load_builtin_engines(); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~ There were also warnings about an infinite recursion error on the header files because of memcmp, memcpy and memset. This is puzzling to me since the kernel compiled just fine in koji. Did it use different header files? And aren't all header files protected from being read more than once? Or are Fedora kernels compiled with warnings as errors turned off? e.g. ./include/linux/fortify-string.h:269:16: note: in expansion of macro '__underlying_memcmp' 269 | return __underlying_memcmp(p, q, size); | ^~~~~~~~~~~~~~~~~~~ In file included from ./include/linux/string.h:253, from ./include/linux/bitmap.h:11, from ./include/linux/cpumask.h:12, from ./arch/x86/include/asm/cpumask.h:5, from ./arch/x86/include/asm/msr.h:11, from ./arch/x86/include/asm/processor.h:22, from ./arch/x86/include/asm/timex.h:5, from ./include/linux/timex.h:65, from ./include/linux/time32.h:13, from ./include/linux/time.h:60, from ./include/linux/stat.h:19, from ./include/linux/module.h:13, from init/do_mounts.c:2: ./include/linux/fortify-string.h: In function 'strncpy': ./include/linux/fortify-string.h:51:24: warning: infinite recursion detected [-Winfinite-recursion] 51 | __FORTIFY_INLINE char *strncpy(char *p, const char *q, __kernel_size_t size) | ^~~~~~~ When I turned off warnings as errors, the kernel compiled successfully. This is not a problem in the 5.16 series, they compile cleanly with warnings as errors turned on. Anyway, I am attaching the error output from the attempt with warnings as errors.

On Sun, 6 Feb 2022 15:25:28 -0600 Justin Forbes <jforbes(a)redhat.com&gt; wrote: > Always best to check kernel-ark's "include in releases" tag to find > out if anything is required for it to build that haven't been merged > yet because review was not finished: > > https://gitlab.com/cki-project/kernel-ark/-/merge_requests?scope=all&... > > There are 2 patches required to properly build with gcc 12, both have > been submitted upstream. Thanks for both the information and the link. I'm not completely clear on what is happening, but I think you added the patches manually to build the 5.17 rc2 Fedora kernel so they are not in the src.rpm. I would have to add them to the spec myself for the kernel 5.17 rc2 to build properly with gcc 12. And once they are accepted upstream, they will be included in the src.rpm. Is that correct?

On Mon, 7 Feb 2022 07:42:54 -0600 Justin Forbes <jforbes(a)redhat.com&gt; wrote: > Oh, no, the patches are in the src rpm and dist-git, they just are > not in the os-build branch of kernel-ark yet, which is what most > people building custom kernels are working with. Apologies, I did > not know you were working from the srpm. I just tried with the latest 5.17, the rc3 src.rpm. And it had all the same errors as the rc2 src.rpm. If the patches are applied, why is the build still showing infinite recursion errors? And it still complains that libperf-jvmti.so is missing when it wants to install it into an rpm. I don't think the patches are in the source. Is there something else I need to do?