On Wed, 31 Aug 2016, Andrew Morton wrote:

quoted

Attempting to online memory which is already online will cause this:

1. store_mem_state() called with buf="online"
2. device_online() returns 1 because device is already online
3. store_mem_state() returns 1
4. calling code interprets this as 1-byte buffer read
5. store_mem_state() called again with buf="nline"
6. store_mem_state() returns -EINVAL

Example:

$ cat /sys/devices/system/memory/memory0/state
online
$ echo online > /sys/devices/system/memory/memory0/state
-bash: echo: write error: Invalid argument

Fix the return value of store_mem_state() so this doesn't happen.

So..  what *does* happen after the patch?  Is some sort of failure still
reported?  Or am I correct in believing that the operation will appear
to have succeeded?  If so, is that desirable?

It's not desirable, before commit 4f3549d72 this would have returned 
EINVAL since __memory_block_change_state() does not see the state as 
MEM_OFFLINE when the write is done.  The correct fix is for 
store_mem_state() to return -EINVAL when device_online() returns non-zero.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>