Date: Wed, 17 Jul 2024 16:17:00 +1200
From: Tao Liu <ltao@redhat.com>
Subject: [Crash-utility] [PATCH v2] Fix "irq -a" exceeding the memory
        range issue
To: devel@lists.crash-utility.osci.io
Cc: Tao Liu <ltao@redhat.com>
Message-ID: <20240717041659.14855-1-ltao@redhat.com>
Content-Type: text/plain; charset="US-ASCII"; x-default=true

Previously without the patch, there was an error observed as follows:

crash> irq -a
IRQ NAME                 AFFINITY
  0 timer                0-191
  4 ttyS0                0-23,96-119
...
 84 smartpqi             72-73,168
irq: page excluded: kernel virtual address: ffff97d03ffff000  type: "irq_desc affinity"

The reason is the reading of irq affinity exceeded the memory range, see
the following debug info:

Thread 1 "crash" hit Breakpoint 1, generic_get_irq_affinity (irq=85) at kernel.c:7373
7375            irq_desc_addr = get_irq_desc_addr(irq);
(gdb) p/x irq_desc_addr
$1 = 0xffff97d03f21e800

crash> struct irq_desc 0xffff97d03f21e800
struct irq_desc {
  irq_common_data = {
    state_use_accessors = 425755136,
    node = 3,
    handler_data = 0x0,
    msi_desc = 0xffff97ca51b83480,
    affinity = 0xffff97d03fffee60,
    effective_affinity = 0xffff97d03fffe6c0
  },

crash> whatis cpumask_t
typedef struct cpumask {
    unsigned long bits[128];
} cpumask_t;
SIZE: 1024

In order to get the affinity, crash will read the memory range 0xffff97d03fffee60
~ 0xffff97d03fffee60 + 1024(0x400) by line:

        readmem(affinity_ptr, KVADDR, affinity, len,
                "irq_desc affinity", FAULT_ON_ERROR);

However the reading will exceed the effective memory range:

crash> kmem 0xffff97d03fffee60
CACHE             OBJSIZE  ALLOCATED     TOTAL  SLABS  SSIZE  NAME
ffff97c900044400       32     123297    162944   1273     4k  kmalloc-32
  SLAB              MEMORY            NODE  TOTAL  ALLOCATED  FREE
  fffffca460ffff80  ffff97d03fffe000     3    128         81    47
  FREE / [ALLOCATED]
  [ffff97d03fffee60]

      PAGE        PHYSICAL      MAPPING       INDEX CNT FLAGS
fffffca460ffff80 83fffe000 dead000000000001 ffff97d03fffe340  1 d7ffffe0000800 slab

crash> kmem ffff97d03ffff000
      PAGE        PHYSICAL      MAPPING       INDEX CNT FLAGS
fffffca460ffffc0 83ffff000                0        0  1 d7ffffe0004000 reserved

crash> dmesg
...
[    0.000000] BIOS-e820: [mem 0x00000000fe000000-0x00000000fe00ffff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000100000000-0x000000083fffefff] usable
[    0.000000] BIOS-e820: [mem 0x000000083ffff000-0x000000083fffffff] reserved
...

The beginning physical address, aka 0x83fffe000, is located in the usable
area and is readable, however the later physical address, starting from
0x83ffff000, is located in reserved region and not readable. In fact,
the affinity member is allocated by alloc_cpumask_var_node(), for the 192 CPUs
system, the allocated size is only 24, and we can see it is within
the kmalloc-32 slab. So it is incorrect to read 1024 length(given by
STRUCT_SIZE("cpumask_t")), only 24 is enough.

Since there are plenty of places in crash which takes the value of
STRUCT_SIZE("cpumask_t"), and works fine for the past, this patch will
not modify them all, only the one which encountered this issue(hunk in
kernel.c), and the one with the same DIV_ROUND_UP() (hunk in tools.c).

Signed-off-by: Tao Liu <ltao@redhat.com>
---
v1 -> v2: modify the same hunk in tools.c

---
 kernel.c |  9 ++++++---
 tools.c  | 11 ++++++++---
 2 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/kernel.c b/kernel.c
index 8a9d498..464e877 100644
--- a/kernel.c
+++ b/kernel.c
@@ -7362,7 +7362,7 @@ void
 generic_get_irq_affinity(int irq)
 {
        ulong irq_desc_addr;
-       long len;
+       long len, len_cpumask;
        ulong affinity_ptr;
        ulong *affinity;
        ulong tmp_addr;
@@ -7382,8 +7382,11 @@ generic_get_irq_affinity(int irq)
        if (!action)
                return;

-       if ((len = STRUCT_SIZE("cpumask_t")) < 0)
-               len = DIV_ROUND_UP(kt->cpus, BITS_PER_LONG) * sizeof(ulong);
+       len = DIV_ROUND_UP(kt->cpus, BITS_PER_LONG) * sizeof(ulong);
+       len_cpumask = STRUCT_SIZE("cpumask_t");
+       if (len_cpumask > 0) {
+               len = len_cpumask > len ? len : len_cpumask;
+       }

        affinity = (ulong *)GETBUF(len);
        if (VALID_MEMBER(irq_common_data_affinity))
diff --git a/tools.c b/tools.c
index 1022d57..fab0f83 100644
--- a/tools.c
+++ b/tools.c
@@ -6718,9 +6718,14 @@ swap64(uint64_t val, int swap)
 ulong *
 get_cpumask_buf(void)
 {
-       int cpulen;
-       if ((cpulen = STRUCT_SIZE("cpumask_t")) < 0)
-               cpulen = DIV_ROUND_UP(kt->cpus, BITS_PER_LONG) * sizeof(ulong);
+       int cpulen, len_cpumask;
+
+       cpulen = DIV_ROUND_UP(kt->cpus, BITS_PER_LONG) * sizeof(ulong);
+       len_cpumask = STRUCT_SIZE("cpumask_t");
+       if (len_cpumask > 0) {
+               cpulen = len_cpumask > cpulen ? cpulen : len_cpumask;
+       }
+
        return (ulong *)GETBUF(cpulen);
 }

--
2.40.1