summaryrefslogtreecommitdiff
path: root/Documentation/arm/mem_alignment
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@ppc970.osdl.org>2005-04-17 02:20:36 +0400
committerLinus Torvalds <torvalds@ppc970.osdl.org>2005-04-17 02:20:36 +0400
commit1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch)
tree0bba044c4ce775e45a88a51686b5d9f90697ea9d /Documentation/arm/mem_alignment
downloadlinux-1da177e4c3f41524e886b7f1b8a0c1fc7321cac2.tar.xz
Linux-2.6.12-rc2
Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
Diffstat (limited to 'Documentation/arm/mem_alignment')
-rw-r--r--Documentation/arm/mem_alignment58
1 files changed, 58 insertions, 0 deletions
diff --git a/Documentation/arm/mem_alignment b/Documentation/arm/mem_alignment
new file mode 100644
index 000000000000..d145ccca169a
--- /dev/null
+++ b/Documentation/arm/mem_alignment
@@ -0,0 +1,58 @@
+Too many problems poped up because of unnoticed misaligned memory access in
+kernel code lately. Therefore the alignment fixup is now unconditionally
+configured in for SA11x0 based targets. According to Alan Cox, this is a
+bad idea to configure it out, but Russell King has some good reasons for
+doing so on some f***ed up ARM architectures like the EBSA110. However
+this is not the case on many design I'm aware of, like all SA11x0 based
+ones.
+
+Of course this is a bad idea to rely on the alignment trap to perform
+unaligned memory access in general. If those access are predictable, you
+are better to use the macros provided by include/asm/unaligned.h. The
+alignment trap can fixup misaligned access for the exception cases, but at
+a high performance cost. It better be rare.
+
+Now for user space applications, it is possible to configure the alignment
+trap to SIGBUS any code performing unaligned access (good for debugging bad
+code), or even fixup the access by software like for kernel code. The later
+mode isn't recommended for performance reasons (just think about the
+floating point emulation that works about the same way). Fix your code
+instead!
+
+Please note that randomly changing the behaviour without good thought is
+real bad - it changes the behaviour of all unaligned instructions in user
+space, and might cause programs to fail unexpectedly.
+
+To change the alignment trap behavior, simply echo a number into
+/proc/sys/debug/alignment. The number is made up from various bits:
+
+bit behavior when set
+--- -----------------
+
+0 A user process performing an unaligned memory access
+ will cause the kernel to print a message indicating
+ process name, pid, pc, instruction, address, and the
+ fault code.
+
+1 The kernel will attempt to fix up the user process
+ performing the unaligned access. This is of course
+ slow (think about the floating point emulator) and
+ not recommended for production use.
+
+2 The kernel will send a SIGBUS signal to the user process
+ performing the unaligned access.
+
+Note that not all combinations are supported - only values 0 through 5.
+(6 and 7 don't make sense).
+
+For example, the following will turn on the warnings, but without
+fixing up or sending SIGBUS signals:
+
+ echo 1 > /proc/sys/debug/alignment
+
+You can also read the content of the same file to get statistical
+information on unaligned access occurrences plus the current mode of
+operation for user space code.
+
+
+Nicolas Pitre, Mar 13, 2001. Modified Russell King, Nov 30, 2001.