Skip to content

Commit

Permalink
drivers/firmware/memmap.c: don't allocate firmware_map_entry of same …
Browse files Browse the repository at this point in the history
…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: sysfs_open_file+0x46/0x2b0
  PGD 203a89fe067 PUD 0
  Oops: 0000 [#1] SMP
  ...
  Call Trace:
    do_dentry_open+0x1ef/0x2a0
    finish_open+0x31/0x40
    do_last+0x57c/0x1220
    path_openat+0xc2/0x4c0
    do_filp_open+0x4b/0xb0
    do_sys_open+0xf3/0x1f0
    SyS_open+0x1e/0x20
    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>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
  • Loading branch information
Yasuaki Ishimatsu authored and torvalds committed Aug 7, 2014
1 parent 49c8b24 commit f0093ed
Showing 1 changed file with 4 additions and 0 deletions.
4 changes: 4 additions & 0 deletions drivers/firmware/memmap.c
Original file line number Diff line number Diff line change
Expand Up @@ -286,6 +286,10 @@ int __meminit firmware_map_add_hotplug(u64 start, u64 end, const char *type)
{
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);
Expand Down

0 comments on commit f0093ed

Please sign in to comment.