作者
查看percpu变量在每个cpu上的基地址
crash> kmem -o
PER-CPU OFFSET VALUES:
CPU 0: ffff88807f600000
CPU 1: ffff88807fa00000
CPU 2: ffff88813d600000
CPU 3: ffff88813da00000
CPU 4: ffff8881bd600000
CPU 5: ffff8881bda00000
CPU 6: ffff88823d600000
CPU 7: ffff88823da00000
查看一个全局的percpu变量的具体值
以下面这个全局percpu变量call_single_queue为例:
static DEFINE_PER_CPU_SHARED_ALIGNED(struct llist_head, call_single_queue);
- 查看这个percpu变量在每个cpu上的地址
crash> p call_single_queue
PER-CPU DATA TYPE:
struct llist_head call_single_queue;
PER-CPU ADDRESSES:
[0]: ffff88807f800340
[1]: ffff88807fc00340
[2]: ffff88813d800340
[3]: ffff88813dc00340
[4]: ffff8881bd800340
[5]: ffff8881bdc00340
[6]: ffff88823d800340
[7]: ffff88823dc00340
- 查看其在某些CPU上的具体内容
crash> p call_single_queue:0,2,5-7
per_cpu(call_single_queue, 0) = $5 = {
first = 0x0
}
per_cpu(call_single_queue, 2) = $6 = {
first = 0x0
}
per_cpu(call_single_queue, 5) = $7 = {
first = 0x0
}
per_cpu(call_single_queue, 6) = $8 = {
first = 0x0
}
per_cpu(call_single_queue, 7) = $9 = {
first = 0x0
}
根据结构体中percpu变量的偏移地址得到绝对地址
以kmem_cache结构体的成员cpu_cache为例:
struct kmem_cache {
struct array_cache __percpu *cpu_cache;
/* 1) Cache tunables. Protected by slab_mutex */
unsigned int batchcount;
unsigned int limit;
unsigned int shared;
...
- 查看偏移量
crash> struct kmem_cache.cpu_slab -x ffff893751f60800
cpu_slab = 0x5fc135c77b40,
上面cpu_slab的偏移量是0x5fc135c77b40
假如想知道这个成员在cpu10上的地址,下面有两种方法:
- 绝对值相加
首先获取percpu变量在cpu10上的基地址:
crash> kmem -o | grep "CPU 10"
CPU 10: ffff88debfd00000
然后相加即可:
crash> eval ffff88debfd00000 + 0x5fc135c77b40
hexadecimal: ffffe89ff5977b40
decimal: 18446718372450630464 (-25701258921152)
octal: 1777777211776545675500
binary: 1111111111111111111010001001111111110101100101110111101101000000
上面将相加后的结果分别按16进制,10进制,8进制以及2进制进行了输出。
- 使用ptov
crash> ptov 0x5fc135c77b40:10
PER-CPU OFFSET: 5fc135c77b40
CPU VIRTUAL
[10] ffffe89ff5977b40
这种方法更加方便。
标签:crash,percpu,CPU,用法,queue,single,call,cpu From: https://www.cnblogs.com/pengdonglin137/p/17724719.html