The kernel says here http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/kernel/events/ring_buffer.c?id=7a1dcf6ad#n61 :
* Since the mmap() consumer (userspace) can run on a different CPU:
*
* kernel user
*
* if (LOAD ->data_tail) { LOAD ->data_head
* (A) smp_rmb() (C)
* STORE $data LOAD $data
* smp_wmb() (B) smp_mb() (D)
* STORE ->data_head STORE ->data_tail
* }
*
* Where A pairs with D, and B pairs with C.
*
* In our case (A) is a control dependency that separates the load of
* the ->data_tail and the stores of $data. In case ->data_tail
* indicates there is no room in the buffer to store $data we do not.
*
* D needs to be a full barrier since it separates the data READ
* from the tail WRITE.
*
* For B a WMB is sufficient since it separates two WRITEs, and for C
* an RMB is sufficient since it separates two READs.
*
* See perf_output_begin().
So I'm pretty sure we need a full barrier before writing out data_tail.
If we check the UID of the pid from /proc/pid/status and the UID matches
our current user, we should be able to forego the authorization dialog.
This should fix a regression where sysprof worked from jhbuild when
profiling a process you have access to.
This fixes the "auto-stop" feature when the child exits, at least in the
case of /usr/bin/gnome-calculator. However, when testing with gedit I
did not see it accurately clean up, so something still needs to be
troubleshooted.
So changing this was a little backwards. The goal here is to allow using
libsysprof-2 without polkit, it just won't have the ability to elevate
privileges.
Just a quick workaround to use a tmpfile if memfd is not available. We
should try to ensure that we get something on tmpfs, but this is probably
good enough for a fallback.
We can use this to notify the user that not enough samples where collected
to build a callgraph. This can happen when all the samples collected are
ignored (due to being in ignored kernel space address ranges for example).
We keep the start time as a iso8601 string, which is useful, but not
enough for relative timestamps to the epoch (start time). Include a
clock time in the same clock domain.
We were not getting the proper value for time. We need sample_id_all set
according to my cursory reading of the core.c in the kernel. Also, the
time is aligned to 64-bit (after the comm field).
