include/linux/mm.h: add new inline function vmf_error()

Many places in drivers/ file systems, error was handled in a common way
like below:

	ret = (ret == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS;

vmf_error() will replace this and return vm_fault_t type err.

A lot of drivers and filesystems currently have a rather complex mapping
of errno-to-VM_FAULT code.  We have been able to eliminate a lot of it
by just returning VM_FAULT codes directly from functions which are
called exclusively from the fault handling path.

Some functions can be called both from the fault handler and other
context which are expecting an errno, so they have to continue to return
an errno.  Some users still need to choose different behaviour for
different errnos, but vmf_error() captures the essential error
translation that's common to all users, and those that need to handle
additional errors can handle them first.

Link: http://lkml.kernel.org/r/20180510174826.GA14268@jordon-HP-15-Notebook-PC
Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/include/linux/mm.h b/include/linux/mm.h
index c080af5..c6fa9a2 100644 (file)
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2466,6 +2466,13 @@ static inline vm_fault_t vmf_insert_pfn(struct vm_area_struct *vma,
        return VM_FAULT_NOPAGE;
 }
 
+static inline vm_fault_t vmf_error(int err)
+{
+       if (err == -ENOMEM)
+               return VM_FAULT_OOM;
+       return VM_FAULT_SIGBUS;
+}
+
 struct page *follow_page_mask(struct vm_area_struct *vma,
                              unsigned long address, unsigned int foll_flags,
                              unsigned int *page_mask);