linux.git/kernel/gcov/Kconfig, branch v5.10.258

gcov: add support for GCC 10.1

2020-09-11T16:33:54+00:00

Using gcov to collect coverage data for kernels compiled with GCC 10.1
causes random malfunctions and kernel crashes.  This is the result of a
changed GCOV_COUNTERS value in GCC 10.1 that causes a mismatch between
the layout of the gcov_info structure created by GCC profiling code and
the related structure used by the kernel.

Fix this by updating the in-kernel GCOV_COUNTERS value.  Also re-enable
config GCOV_KERNEL for use with GCC 10.

Reported-by: Colin Ian King 
Reported-by: Leon Romanovsky 
Signed-off-by: Peter Oberparleiter 
Tested-by: Leon Romanovsky 
Tested-and-Acked-by: Colin Ian King 
Signed-off-by: Linus Torvalds

gcov: Disable gcov build with GCC 10

2020-09-04T16:19:49+00:00

GCOV built with GCC 10 doesn't initialize n_function variable.  This
produces different kernel panics as was seen by Colin in Ubuntu and me
in FC 32.

As a workaround, let's disable GCOV build for broken GCC 10 version.

Link: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1891288
Link: https://lore.kernel.org/lkml/20200827133932.3338519-1-leon@kernel.org
Link: https://lore.kernel.org/lkml/CAHk-=whbijeSdSvx-Xcr0DPMj0BiwhJ+uiNnDSVZcr_h_kg7UA@mail.gmail.com/
Cc: Colin Ian King 
Signed-off-by: Leon Romanovsky 
Signed-off-by: Linus Torvalds

treewide: replace '---help---' in Kconfig files with 'help'

2020-06-13T16:57:21+00:00

Since commit 84af7a6194e4 ("checkpatch: kconfig: prefer 'help' over
'---help---'"), the number of '---help---' has been gradually
decreasing, but there are still more than 2400 instances.

This commit finishes the conversion. While I touched the lines,
I also fixed the indentation.

There are a variety of indentation styles found.

  a) 4 spaces + '---help---'
  b) 7 spaces + '---help---'
  c) 8 spaces + '---help---'
  d) 1 space + 1 tab + '---help---'
  e) 1 tab + '---help---'    (correct indentation)
  f) 1 tab + 1 space + '---help---'
  g) 1 tab + 2 spaces + '---help---'

In order to convert all of them to 1 tab + 'help', I ran the
following commend:

  $ find . -name 'Kconfig*' | xargs sed -i 's/^[[:space:]]*---help---/\thelp/'

Signed-off-by: Masahiro Yamada

gcov: Remove old GCC 3.4 support

2020-04-16T11:28:35+00:00

The kernel requires at least GCC 4.8 in order to build, and so there is
no need to cater for the pre-4.7 gcov format.

Remove the obsolete code.

Acked-by: Peter Oberparleiter 
Reviewed-by: Nick Desaulniers 
Signed-off-by: Will Deacon

Revert "um: Enable CONFIG_CONSTRUCTORS"

2020-01-19T21:42:06+00:00

This reverts commit 786b2384bf1c ("um: Enable CONFIG_CONSTRUCTORS").

There are two issues with this commit, uncovered by Anton in tests
on some (Debian) systems:

1) I completely forgot to call any constructors if CONFIG_CONSTRUCTORS
   isn't set. Don't recall now if it just wasn't needed on my system, or
   if I never tested this case.

2) With that fixed, it works - with CONFIG_CONSTRUCTORS *unset*. If I
   set CONFIG_CONSTRUCTORS, it fails again, which isn't totally
   unexpected since whatever wanted to run is likely to have to run
   before the kernel init etc. that calls the constructors in this case.

Basically, some constructors that gcc emits (libc has?) need to run
very early during init; the failure mode otherwise was that the ptrace
fork test already failed:

----------------------
$ ./linux mem=512M
Core dump limits :
	soft - 0
	hard - NONE
Checking that ptrace can change system call numbers...check_ptrace : child exited with exitcode 6, while expecting 0; status 0x67f
Aborted
----------------------

Thinking more about this, it's clear that we simply cannot support
CONFIG_CONSTRUCTORS in UML. All the cases we need now (gcov, kasan)
involve not use of the __attribute__((constructor)), but instead
some constructor code/entry generated by gcc. Therefore, we cannot
distinguish between kernel constructors and system constructors.

Thus, revert this commit.

Cc: stable@vger.kernel.org [5.4+]
Fixes: 786b2384bf1c ("um: Enable CONFIG_CONSTRUCTORS")
Reported-by: Anton Ivanov 
Signed-off-by: Johannes Berg 
Acked-by: Anton Ivanov 

Signed-off-by: Richard Weinberger

um: Enable CONFIG_CONSTRUCTORS

2019-09-15T19:37:13+00:00

We do need to call the constructors for *modules*, and
at least for KASAN in the future, we must call even the
kernel constructors only later when the kernel has been
initialized.

Instead of relying on libc to call them, emit an empty
section for libc and let the kernel's CONSTRUCTORS code
do the rest of the job.

Tested that it indeed doesn't work in modules, and does
work after the fixes in both, with a few functions with
__attribute__((constructor)) in both dynamic and static
builds.

Signed-off-by: Johannes Berg 
Signed-off-by: Richard Weinberger

treewide: Add SPDX license identifier - Makefile/Kconfig

2019-05-21T08:50:46+00:00

Add SPDX license identifiers to all Make/Kconfig files which:

 - Have no license information of any form

These files fall under the project license, GPL v2 only. The resulting SPDX
license identifier is:

  GPL-2.0-only

Signed-off-by: Thomas Gleixner 
Signed-off-by: Greg Kroah-Hartman

gcov: clang support

2019-05-15T02:52:51+00:00

LLVM uses profiling data that's deliberately similar to GCC, but has a
very different way of exporting that data.  LLVM calls llvm_gcov_init()
once per module, and provides a couple of callbacks that we can use to
ask for more data.

We care about the "writeout" callback, which in turn calls back into
compiler-rt/this module to dump all the gathered coverage data to disk:

   llvm_gcda_start_file()
     llvm_gcda_emit_function()
     llvm_gcda_emit_arcs()
     llvm_gcda_emit_function()
     llvm_gcda_emit_arcs()
     [... repeats for each function ...]
   llvm_gcda_summary_info()
   llvm_gcda_end_file()

This design is much more stateless and unstructured than gcc's, and is
intended to run at process exit.  This forces us to keep some local
state about which module we're dealing with at the moment.  On the other
hand, it also means we don't depend as much on how LLVM represents
profiling data internally.

See LLVM's lib/Transforms/Instrumentation/GCOVProfiling.cpp for more
details on how this works, particularly GCOVProfiler::emitProfileArcs(),
GCOVProfiler::insertCounterWriteout(), and GCOVProfiler::insertFlush().

[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/20190417225328.208129-1-trong@android.com
Signed-off-by: Greg Hackmann 
Signed-off-by: Nick Desaulniers 
Signed-off-by: Tri Vo 
Co-developed-by: Nick Desaulniers 
Co-developed-by: Tri Vo 
Tested-by: Trilok Soni 
Tested-by: Prasad Sodagudi 
Tested-by: Tri Vo 
Tested-by: Daniel Mentz 
Tested-by: Petri Gynther 
Reviewed-by: Peter Oberparleiter 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

gcov: remove CONFIG_GCOV_FORMAT_AUTODETECT

2018-06-08T09:56:02+00:00

CONFIG_GCOV_FORMAT_AUTODETECT compiles either gcc_3_4.c or gcc_4_7.c
according to your GCC version.

We can achieve the equivalent behavior by setting reasonable dependency
with the knowledge of the compiler version.

If GCC older than 4.7 is used, GCOV_FORMAT_3_4 is the default, but users
are still allowed to select GCOV_FORMAT_4_7 in case the newer format is
back-ported.

On the other hand, If GCC 4.7 or newer is used, there is no reason to
use GCOV_FORMAT_3_4, so it should be hidden.

If you downgrade the compiler to GCC 4.7 or older, oldconfig/syncconfig
will display a prompt for the choice because GCOV_FORMAT_3_4 becomes
visible as a new symbol.

Signed-off-by: Masahiro Yamada 
Acked-by: Peter Oberparleiter 
Reviewed-by: Kees Cook

gcov: disable for COMPILE_TEST

2016-05-10T15:12:49+00:00

Enabling gcov is counterproductive to compile testing: it significantly
increases the kernel image size, compile time, and it produces lots
of false positive "may be used uninitialized" warnings as the result
of missed optimizations.

This is in line with how UBSAN_SANITIZE_ALL and PROFILE_ALL_BRANCHES
work, both of which have similar problems.

With an ARM allmodconfig kernel, I see the build time drop from
283 minutes CPU time to 225 minutes, and the vmlinux size drops
from 43MB to 26MB.

Signed-off-by: Arnd Bergmann 
Acked-by: Peter Oberparleiter 
Signed-off-by: Michal Marek