怎么观察一个用户态程序的内核API调用情况

simtiger

比如一个用户态程序在反复申请、释放内存，，，，我想通过某个命令或者某个方法查看当前系统的内核API的调用情况，比如：
想看到下面类似的信息：
brk(0xc5673b000)                         = 0xffffffffc3527000
mmap2(NULL, 123985, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0xffffb5c3) = -1 ENOMEM (Cannot allocate memory) mmap2(NULL, 2023152, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, -1, 0xffffb254) = -1 ENOMEM (Cannot allocate memory) mmap2(NULL, 1048326, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, -1, 0xffffb254) = -1 ENOMEM (Cannot allocate memory) mmap2(NULL, 2027152, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, -1, 0xffffb254) = -1 ENOMEM (Cannot allocate memory) mmap2(NULL, 1038576, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, -1, 0xffffb254) = -1 ENOMEM (Cannot allocate memory)
....

这样的信息应该如何查看啊？请教一下。

[ 本帖最后由 simtiger 于 2009-8-18 10:23 编辑 ]

kouu

strace

iknowabc

truss/strace/ltrace都可以，其实他们都是基于ptrace这个系统调用(attach到目标进程中的)
truss -o output.txt xxx
strace -f -o output.txt xxx
ltrace -p pid

xgr180

ftrace 中的function/function_graph tracer