I got following GCOV configs enabled
CONFIG_GCOV_KERNEL=y
CONFIG_ARCH_HAS_GCOV_PROFILE_ALL=y
CONFIG_GCOV_PROFILE_ALL=y
and was trying to run UML (kernel version 6.1.44) and ended up in a
crash in printk_caller_id().
I tried to debug and it seems current_thread_info is returning NULL
and that is because it is not yet set up. It hit much before mm_init
or any thread_info could be created.
this print is actually coming from __gcov_init
void __gcov_init(struct gcov_info *info)
{
static unsigned int gcov_version;
mutex_lock(&gcov_lock);
if (gcov_version == 0) {
gcov_version = gcov_info_version(info);
/*
* Printing gcc's version magic may prove useful for debugging
* incompatibility reports.
*/
// Need to uncomment post fixing NC-124446
pr_info("version magic: 0x%x\n", gcov_version);
<<<<<<< this pr_info causing crash
$ gdb ./linux
(gdb) run mem=512M rootfstype=hostfs rw init=/bin/bash
Starting program:
/home/vimal.agrawal/Morane/git-openwrt/basesystem/NEMO/linux-kernel/linux
mem=512M rootfstype=hostfs rw init=/bin/bash
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Program received signal SIGSEGV, Segmentation fault.
0x00000000600f2de3 in printk_caller_id () at kernel/printk/printk.c:2033
2033 return in_task() ? task_pid_nr(current) :
(gdb) bt
#0 0x00000000600f2de3 in printk_caller_id () at kernel/printk/printk.c:2033
#1 vprintk_store (facility=0, level=<optimized out>, dev_info=0x0,
fmt=0x6097ed2a "\001\066version magic: 0x%x\n",
args=0x7fffffffe188) at kernel/printk/printk.c:2143
#2 0x0000000000000000 in ?? ()
I am not sure if any of my .config is causing this. I don't think we
can use pr_info so early. I do see that this pr_info in __gcov_init
was added long back in 2019.
Thanks,
Vimal
_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um
