+ dont-allocate-firmware_map_entry-of-same-memory-range.patch added to -mm tree

+ dont-allocate-firmware_map_entry-of-same-memory-range.patch added to -mm tree

[akpm]

The patch titled
     Subject: drivers/firmware/memmap.c: don't allocate firmware_map_entry of same memory range
has been added to the -mm tree.  Its filename is
     dont-allocate-firmware_map_entry-of-same-memory-range.patch

This patch should soon appear at
    http://ozlabs.org/~akpm/mmots/broken-out/dont-allocate-firmware_map_entry-of-same-memory-range.patch
and later at
    http://ozlabs.org/~akpm/mmotm/broken-out/dont-allocate-firmware_map_entry-of-same-memory-range.patch

Before you just go and hit "reply", please:
   a) Consider who else should be cc'ed
   b) Prefer to cc a suitable mailing list as well
   c) Ideally: find the original patch on the mailing list and do a
      reply-to-all to that, adding suitable additional cc's

*** Remember to use Documentation/SubmitChecklist when testing your code ***

The -mm tree is included into linux-next and is updated
there every 3-4 working days

------------------------------------------------------
From: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Subject: drivers/firmware/memmap.c: don't allocate firmware_map_entry of same memory range

When limiting memory by mem= and ACPI DSDT table has PNP0C80,
firmware_map_entrys of same memory range are allocated and memmap X
sysfses which have same memory range are created as follows:

 # cat /sys/firmware/memmap/0/*
 0x407ffffffff
 0x40000000000
 System RAM
 # cat /sys/firmware/memmap/33/*
 0x407ffffffff
 0x40000000000
 System RAM
 # cat /sys/firmware/memmap/35/*
 0x407ffffffff
 0x40000000000
 System RAM

In this case, when hot-removing memory, kernel panic occurs, showing
following call trace:

 BUG: unable to handle kernel paging request at 00000001003e000b
 IP: [<ffffffff81225c26>] sysfs_open_file+0x46/0x2b0
 PGD 203a89fe067 PUD 0
 Oops: 0000 [#1] SMP
 ...
 Call Trace:
  [<ffffffff811ad95f>] do_dentry_open+0x1ef/0x2a0
  [<ffffffff811bba32>] ? __inode_permission+0x52/0xc0
  [<ffffffff81225be0>] ? sysfs_schedule_callback+0x1c0/0x1c0
  [<ffffffff811ada41>] finish_open+0x31/0x40
  [<ffffffff811bf6ac>] do_last+0x57c/0x1220
  [<ffffffff8119529e>] ? kmem_cache_alloc_trace+0x1ce/0x1f0
  [<ffffffff811c0412>] path_openat+0xc2/0x4c0
  [<ffffffff81167e2c>] ? tlb_flush_mmu.part.53+0x4c/0x90
  [<ffffffff811c102b>] do_filp_open+0x4b/0xb0
  [<ffffffff811cd827>] ? __alloc_fd+0xa7/0x130
  [<ffffffff811aeed3>] do_sys_open+0xf3/0x1f0
  [<ffffffff811aefee>] SyS_open+0x1e/0x20
  [<ffffffff815f2199>] system_call_fastpath+0x16/0x1b

The problem occurs as follows:

When calling e820_reserve_resources(), firmware_map_entrys of all
e820 memory map are allocated. And all firmware_map_entrys is added
map_entries list as follows:

map_entries
 -> +--- entry A --------+ -> ...
    | start 0x407ffffffff|
    | end   0x40000000000|
    | type  System RAM   |
    +--------------------+

After that, if ACPI DSDT table has PNP0C80 and the memory range is
limited by mem=, the PNP0C80 is hot-added. Then firmware_map_entry of
PNP0C80 is allocated and added map_entries list as follows:

map_entries
 -> +--- entry A --------+ -> ... -> +--- entry B --------+
    | start 0x407ffffffff|           | start 0x407ffffffff|
    | end   0x40000000000|           | end   0x40000000000|
    | type  System RAM   |           | type  System RAM   |
    +--------------------+           +--------------------+

Then memmap 0 sysfs for entry B is created.

After that, firmware_memmap_init() creates memmap sysfses of all
firmware_map_entrys in map_entries list.  As a result, memmap 33 sysfs for
entry A and memmap 35 sysfs for entry B are created.  But kobject of entry
B has been used by memmap 0 sysfs.  So when creating memmap 35 sysfs, the
kobject is broken.

If hot-removing memory, memmap 0 sysfs is destroyed and kobject of memmap
0 sysfs is freed.  But the kobject can be accessed via memmap 35 sysfs. 
So when open memmap 35 sysfs, kernel panic occurs.

This patch checks whether there is firmware_map_entry of same memory range
in map_entries list and don't allocate firmware_map_entry of same memroy
range.

Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 drivers/firmware/memmap.c |    4 ++++
 1 file changed, 4 insertions(+)

diff -puN drivers/firmware/memmap.c~dont-allocate-firmware_map_entry-of-same-memory-range drivers/firmware/memmap.c
--- a/drivers/firmware/memmap.c~dont-allocate-firmware_map_entry-of-same-memory-range
+++ a/drivers/firmware/memmap.c
@@ -286,6 +286,10 @@ int __meminit firmware_map_add_hotplug(u
 {
 	struct firmware_map_entry *entry;
 
+	entry = firmware_map_find_entry(start, end - 1, type);
+	if (entry)
+		return 0;
+
 	entry = firmware_map_find_entry_bootmem(start, end - 1, type);
 	if (!entry) {
 		entry = kzalloc(sizeof(struct firmware_map_entry), GFP_ATOMIC);
_

Patches currently in -mm which might be from isimatu.yasuaki@jp.fujitsu.com are

x86mem-hotplug-pass-sync_global_pgds-a-correct-argument-in-remove_pagetable.patch
x86mem-hotplug-modify-pgd-entry-when-removing-memory.patch
x86-numa-setup_node_data-drop-dead-code-and-rename-function.patch
x86-numa-setup_node_data-drop-dead-code-and-rename-function-checkpatch-fixes.patch
mem-hotplug-avoid-illegal-state-prefixed-with-legal-state-when-changing-state-of-memory_block.patch
mem-hotplug-introduce-mmop_offline-to-replace-the-hard-coding-1.patch
firmware-memmap-pass-the-correct-argument-to-firmware_map_find_entry_bootmem.patch
dont-allocate-firmware_map_entry-of-same-memory-range.patch