Source: rdma-core Version: 15-1 Severity: important Tags: upstream Justification: fails to build from source User: debian-m...@lists.debian.org Usertags: mips mips64el mipsel
Builds of rdma-core for armhf and mips* have been failing: -- Performing Test HAVE_COHERENT_DMA -- Performing Test HAVE_COHERENT_DMA - Failed [...] -- Missing Optional Items: -- Compiler attribute ifunc NOT supported -- Architecture NOT able to do coherent DMA (check util/udma_barrier.h) some providers disabled! [...] dh_install --fail-missing -X"etc/init.d/srpd" -X"usr/sbin/run_srp_daemon" -X"usr/sbin/srp_daemon.sh" dh_install: Please use dh_missing --list-missing/--fail-missing instead dh_install: This feature will be removed in compat 12. dh_install: Cannot find (any matches for) "etc/modprobe.d/mlx4.conf" (tried in ., debian/tmp) [...] dh_install: libibverbs-dev missing files: usr/share/man/man7/mlx5dv.7 dh_install: missing files, aborting debian/rules:45: recipe for target 'override_dh_install' failed make[1]: *** [override_dh_install] Error 25 As the log indicates, upstream compensates for udma_barrier.h's limited architecture coverage by disabling providers that need coherent DMA rather than failing the build altogether. As such, one possibility would be to accommodate these providers' potential absence by (e.g.) wildcarding more aggressively. However, it would of course be better to teach udma_barrier.h about these architectures. I don't know what appropriate definitions of udma_to_device_barrier, udma_from_device_barrier, and mmio_flush_writes would be there, but I'm usertagging porters who presumably will know what to use. Could you please take a look? Thanks! -- Aaron M. Ucko, KB1CJC (amu at alum.mit.edu, ucko at debian.org) http://www.mit.edu/~amu/ | http://stuff.mit.edu/cgi/finger/?a...@monk.mit.edu