Re: [RFC PATCH] bug: always show source-tree-relative paths in WARN()/BUG()

2019-07-12 Thread Brian Norris
On Thu, Jul 11, 2019 at 6:50 PM Masahiro Yamada wrote: > GCC 8 added this flag. > So, it will be eventually all solved in the GCC world. Ack. > Clang has not supported it yet... That's what it appeared like. I've bugged our Clang-loving toolchain folks to see if we can get parity. > Trimming a

Re: [RFC PATCH] bug: always show source-tree-relative paths in WARN()/BUG()

2019-07-11 Thread Masahiro Yamada
On Fri, Jul 12, 2019 at 10:23 AM Brian Norris wrote: > > On Thu, Jul 11, 2019 at 6:14 PM Masahiro Yamada > wrote: > > BTW, did you see this? > > > > commit a73619a845d5625079cc1b3b820f44c899618388 > > Author: Masahiro Yamada > > Date: Fri Mar 30 13:15:26 2018 +0900 > > > > kbuild: use -fma

Re: [RFC PATCH] bug: always show source-tree-relative paths in WARN()/BUG()

2019-07-11 Thread Brian Norris
On Thu, Jul 11, 2019 at 6:14 PM Masahiro Yamada wrote: > BTW, did you see this? > > commit a73619a845d5625079cc1b3b820f44c899618388 > Author: Masahiro Yamada > Date: Fri Mar 30 13:15:26 2018 +0900 > > kbuild: use -fmacro-prefix-map to make __FILE__ a relative path Oh, wow, no I did not. If

Re: [RFC PATCH] bug: always show source-tree-relative paths in WARN()/BUG()

2019-07-11 Thread Masahiro Yamada
On Fri, Jul 12, 2019 at 10:06 AM Brian Norris wrote: > > When building out-of-tree (e.g., 'make O=...'), __FILE__ ends up being > an absolute path, and so WARN() and BUG() end up putting path names from > the build system into the log text. For example: > > # echo BUG > /sys/kernel/debug/provoke

[RFC PATCH] bug: always show source-tree-relative paths in WARN()/BUG()

2019-07-11 Thread Brian Norris
When building out-of-tree (e.g., 'make O=...'), __FILE__ ends up being an absolute path, and so WARN() and BUG() end up putting path names from the build system into the log text. For example: # echo BUG > /sys/kernel/debug/provoke-crash/DIRECT ... kernel BUG at /mnt/host/source/[...]/driver