fix include order in mthca_memfree.c
PATCH 1/1 fix include order in mthca_memfree.c
About commit fdca124a1bcc7e624f6b887c6f26153f40ee43ee ("missing include in mthca"):
- Out of curiousity, what arch and/or config requires <linux/mm.h>? I regularly cross-compile drivers/infiniband for i386, x86_64, ppc64, ia64, sparc64 and ppc, and I haven't needed <linux/mm.h> in mthca_memfree.c.
- When making changes to drivers/infiniband, can you please cc the maintainers or at least a public mailing list? As far as I can tell, the patch went directly to Linus with no public review, which doesn't seem appropriate, no matter how trivial the change.
- When adding includes, please match the existing style and put <linux/xxx.h> includes before any local "yyy.h" includes.
Linus, please apply the patch below to move the include where it belongs.
Thanks,
Roland
Fix order of #include lines in mthca_memfree.c
drivers/infiniband/hw/mthca/mthca_memfree.c +2 -1
@@ -32,10 +32,11 @@* $Id$ */ +#include <linux/mm.h> +#include "mthca_memfree.h" #include "mthca_dev.h" #include "mthca_cmd.h" -#include <linux/mm.h> /* * We allocate in as big chunks as we can, up to a maximum of 256 KB
Quoted 3 lines · Roland Dreier
- Out of curiousity, what arch and/or config requires <
linux/mm.h>? I regularly cross-compiledrivers/infinibandfor i386,x86_64, ppc64, ia64, sparc64 and ppc, and I haven't needed <linux/mm.h> inmthca_memfree.c.
alpha, for instance. You are using lowmem_page_address(). That's from linux/mm.h and that's an inline function, so missing include is fatal.
FWIW, the chain of includes that leads to mm.h on i386 is mthca_memfree.h -> linux/pci.h -> asm-i386/pci.h -> linux/mm.h. Other platforms differ...
Quoted 7 lines
- When making changes to
drivers/infiniband, can you please cc the maintainers or at least a public mailing list? As far as I can tell, the patch went directly to Linus with no public review, which doesn't seem appropriate, no matter how trivial the change.- When adding includes, please match the existing style and put <
linux/xxx.h> includes before any local "yyy.h" includes.
Sure, no problem.