问题
I'm trying to compile "slock" to implement some tweaks. It fails to start, on this:
#ifdef __linux__
#include <fcntl.h>
static void
dontkillme(void) {
int fd;
fd = open("/proc/self/oom_score_adj", O_WRONLY);
if (fd < 0 && errno == ENOENT)
return;
if (fd < 0 || write(fd, "-1000\n", 6) != 6 || close(fd) != 0)
die("cannot disable the out-of-memory killer for this process\n");
}
#endif
I went ahead and did some debugging, and found out I get EACCESS from the write().
What's the catch? I believe this is some security feature, as it's not cool if processes can change this value - but what am I doing wrong? Is there some special filesystem flag on the binary to be set, to get this work?
回答1:
Looking at Documentation/filesystems/proc.txt in the kernel:
The value of /proc/pid/oom_score_adj may be reduced no lower than the last value set by a CAP_SYS_RESOURCE process. To reduce the value any lower requires CAP_SYS_RESOURCE.
Which means that you must be either root or have CAP_SYS_RESOURCE capability at some point to set this lower than the default.
来源：https://stackoverflow.com/questions/26963700/eaccess-writing-proc-self-oom-score-adj