Skip to content

L
linux
Switch branch/tag
  1. 10 May, 2019 2 commits
    • Steven Rostedt (VMware)'s avatar
      ftrace/x86: Remove mcount support · 562e14f7
      Steven Rostedt (VMware) authored 
      There's two methods of enabling function tracing in Linux on x86. One is
with just "gcc -pg" and the other is "gcc -pg -mfentry". The former will use
calls to a special function "mcount" after the frame is set up in all C
functions. The latter will add calls to a special function called "fentry"
as the very first instruction of all C functions.

At compile time, there is a check to see if gcc supports, -mfentry, and if
it does, it will use that, because it is more versatile and less error prone
for function tracing.

Starting with v4.19, the minimum gcc supported to build the Linux kernel,
was raised to version 4.6. That also happens to be the first gcc version to
support -mfentry. Since on x86, using gcc versions from 4.6 and beyond will
unconditionally enable the -mfentry, it will no longer use mcount as the
method for inserting calls into the C functions of the kernel. This means
that there is no point in continuing to maintain mcount in x86.

Remove support for using mcount. This makes the code less complex, and will
also allow it to be simplified in the future.
Acked-by: default avatarPeter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: default avatarJiri Kosina <jkosina@suse.cz>
Acked-by: default avatarJosh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: default avatarSteven Rostedt (VMware) <rostedt@goodmis.org>
      562e14f7
    • Steven Rostedt (VMware)'s avatar
      ftrace/x86_32: Remove support for non DYNAMIC_FTRACE · 518049d9
      Steven Rostedt (VMware) authored 
      When DYNAMIC_FTRACE is enabled in the kernel, all the functions that can be
traced by the function tracer have a "nop" placeholder at the start of the
function. When function tracing is enabled, the nop is converted into a call
to the tracing infrastructure where the functions get traced. This also
allows for specifying specific functions to trace, and a lot of
infrastructure is built on top of this.

When DYNAMIC_FTRACE is not enabled, all the functions have a call to the
ftrace trampoline. A check is made to see if a function pointer is the
ftrace_stub or not, and if it is not, it calls the function pointer to trace
the code. This adds over 10% overhead to the kernel even when tracing is
disabled.

When an architecture supports DYNAMIC_FTRACE there really is no reason to
use the static tracing. I have kept non DYNAMIC_FTRACE available in x86 so
that the generic code for non DYNAMIC_FTRACE can be tested. There is no
reason to support non DYNAMIC_FTRACE for both x86_64 and x86_32. As the non
DYNAMIC_FTRACE for x86_32 does not even support fentry, and we want to
remove mcount completely, there's no reason to keep non DYNAMIC_FTRACE
around for x86_32.
Signed-off-by: default avatarSteven Rostedt (VMware) <rostedt@goodmis.org>
      518049d9
  3. 09 May, 2019 1 commit
  5. 08 May, 2019 13 commits
  7. 03 May, 2019 3 commits
  9. 29 Apr, 2019 1 commit
  11. 21 Apr, 2019 1 commit
    • Steven Rostedt (VMware)'s avatar
      function_graph: Have selftest also emulate tr->reset() as it did with tr->init() · 52fde6e7
      Steven Rostedt (VMware) authored 
      The function_graph boot up self test emulates the tr->init() function in
order to add a wrapper around the function graph tracer entry code to test
for lock ups and such. But it does not emulate the tr->reset(), and just
calls the function_graph tracer tr->reset() function which will use its own
fgraph_ops to unregister function tracing with. As the fgraph_ops is
becoming more meaningful with the register_ftrace_graph() and
unregister_ftrace_graph() functions, the two need to be the same. The
emulated tr->init() uses its own fgraph_ops descriptor, which means the
unregister_ftrace_graph() must use the same ftrace_ops, which the selftest
currently does not do. By emulating the tr->reset() as the selftest does
with the tr->init() it will be able to pass the same fgraph_ops descriptor
to the unregister_ftrace_graph() as it did with the register_ftrace_graph().
Signed-off-by: default avatarSteven Rostedt (VMware) <rostedt@goodmis.org>
      52fde6e7
  13. 11 Apr, 2019 1 commit
    • Steven Rostedt (VMware)'s avatar
      ftrace: Do not process STUB functions in ftrace_ops_list_func() · 2fa717a0
      Steven Rostedt (VMware) authored 
      The function_graph tracer has a stub function and its ops flag has the
FTRACE_OPS_FL_STUB set. As the function graph does not use the
ftrace_ops->func pointer but instead is called by a separate part of the
ftrace trampoline. The function_graph tracer still requires to pass in a
ftrace_ops that may also hold the hash of the functions to call. But there's
no reason to test that hash in the function tracing portion. Instead of
testing to see if we should call the stub function, just test if the ops has
FTRACE_OPS_FL_STUB set, and just skip it.
Signed-off-by: default avatarSteven Rostedt (VMware) <rostedt@goodmis.org>
      2fa717a0
  15. 10 Apr, 2019 1 commit
  17. 08 Apr, 2019 12 commits
  19. 02 Apr, 2019 5 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7