linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Tor Vic <torvic9@mailbox.org>
To: Masahiro Yamada <masahiroy@kernel.org>
Cc: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Nathan Chancellor <nathan@kernel.org>,
	"ndesaulniers@google.com" <ndesaulniers@google.com>,
	Kees Cook <keescook@chromium.org>,
	Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
	"clang-built-linux@googlegroups.com" 
	<clang-built-linux@googlegroups.com>
Subject: Re: [PATCH v2 1/1] Kbuild, clang: add option for choosing a ThinLTO cache directory
Date: Wed, 14 Jul 2021 06:12:07 +0000	[thread overview]
Message-ID: <bef18150-3f34-9739-5532-3a3533574fa4@mailbox.org> (raw)
In-Reply-To: <CAK7LNAT9oMkSthmCJ9sq3PFRcYgZoC6O0o==WQnKOU0znzT6hQ@mail.gmail.com>



On 14.07.21 02:46, Masahiro Yamada wrote:
> On Wed, Jul 14, 2021 at 1:41 AM Tor Vic <torvic9@mailbox.org> wrote:
>>
>> On some distros and configurations, it might be useful to allow for
>> specifying a directory where Clang stores its ThinLTO cache.
>>
>> More specifically, when building the VirtualBox extramodules on Arch with
>> its proper 'makepkg' build system and DKMS, against an already installed
>> ThinLTO kernel, the build fails because it tries to create the ThinLTO
>> cache in a directory that is not user-writable.
> 
> 
> Again, I do not understand.

Example:
1. build and install a kernel+headers with Arch's own makepkg/pacman tools.
2. at the end of the install, dkms is automatically called to rebuild
any out-of-tree modules like e.g. VirtualBox.
3. dkms fails because it tries to store the thinlto cache in a
directory that it not user-writable (very possibly in the installed
kernel headers directory under /usr/lib).
4. By explicitly selecting a place for the cache, e.g. /tmp, the dkms
module rebuild succeeds.

> 
> Is this fixing the root cause?

If only I knew. I have very limited skills, and no IT or coding
background. It's possible that this can be fixed also on dkms but
I'm not sure.
Currently, dkms does not do a compiler check (it only seems to know
gcc) and it will fail unless explicitly told to use clang. e.g.
by specifying 'CC=clang LD=ld.lld" etc.
This has been reported to them.

> 
> To me, it looks like
> "Anyway, this worked for me" patch.

At least one other person reported success by moving the cache to
another place (see link).
Admittedly, it looks the same way to me, I just wanted to get things
to work and it's likely that there are better solutions to this
problem, but AFAIR no one came forward.

> 
> Besides, 'make clean' will never clean up the
> cache directory.

I can have a look at it again.

> 
> 
> 
> 
> 
>> A similar problem has been reported with openSUSE's OBS build system.
>>
>> Add a Kconfig option that allows users to choose a directory in which
>> Clang's ThinLTO can store its cache.
>>
>> Link: https://github.com/ClangBuiltLinux/linux/issues/1104
>> Signed-off-by: Tor Vic <torvic9@mailbox.org>
>> ---
>> Changes from v1 to v2: remove unneeded changes in scripts/Makefile
>>
>>  Makefile     |  5 +++--
>>  arch/Kconfig | 10 ++++++++++
>>  2 files changed, 13 insertions(+), 2 deletions(-)
>>
>> diff --git a/Makefile b/Makefile
>> index c3f9bd191b89..472bc8bfff03 100644
>> --- a/Makefile
>> +++ b/Makefile
>> @@ -932,7 +932,8 @@ endif
>>  ifdef CONFIG_LTO_CLANG
>>  ifdef CONFIG_LTO_CLANG_THIN
>>  CC_FLAGS_LTO   := -flto=thin -fsplit-lto-unit
>> -KBUILD_LDFLAGS += --thinlto-cache-dir=$(extmod_prefix).thinlto-cache
>> +export thinlto-dir = $(if
>> $(CONFIG_LTO_CLANG_THIN_CACHEDIR),$(CONFIG_LTO_CLANG_THIN_CACHEDIR)/)
>> +KBUILD_LDFLAGS +=
>> --thinlto-cache-dir=$(thinlto-dir)$(extmod_prefix).thinlto-cache
>>  else
>>  CC_FLAGS_LTO   := -flto
>>  endif
>> @@ -1728,7 +1729,7 @@ PHONY += compile_commands.json
>>
>>  clean-dirs := $(KBUILD_EXTMOD)
>>  clean: rm-files := $(KBUILD_EXTMOD)/Module.symvers
>> $(KBUILD_EXTMOD)/modules.nsdeps \
>> -       $(KBUILD_EXTMOD)/compile_commands.json $(KBUILD_EXTMOD)/.thinlto-cache
>> +       $(KBUILD_EXTMOD)/compile_commands.json
>> $(thinlto-dir)$(KBUILD_EXTMOD)/.thinlto-cache
>>
>>  PHONY += help
>>  help:
>> diff --git a/arch/Kconfig b/arch/Kconfig
>> index 129df498a8e1..19e4d140e12a 100644
>> --- a/arch/Kconfig
>> +++ b/arch/Kconfig
>> @@ -696,6 +696,16 @@ config LTO_CLANG_THIN
>>             https://clang.llvm.org/docs/ThinLTO.html
>>
>>           If unsure, say Y.
>> +
>> +config LTO_CLANG_THIN_CACHEDIR
>> +       string "Clang ThinLTO cache directory"
>> +       depends on LTO_CLANG_THIN
>> +       default ""
>> +       help
>> +         This option allows users to choose a directory that stores
>> +         Clang's ThinLTO cache.
>> +         Leave empty for default.
>> +
>>  endchoice
>>
>>  config ARCH_SUPPORTS_CFI_CLANG
>> --
>> 2.32.0
>>
> 
> 

      reply	other threads:[~2021-07-14  6:18 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-13 16:41 [PATCH v2 1/1] Kbuild, clang: add option for choosing a ThinLTO cache directory Tor Vic
2021-07-14  2:46 ` Masahiro Yamada
2021-07-14  6:12   ` Tor Vic [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=bef18150-3f34-9739-5532-3a3533574fa4@mailbox.org \
    --to=torvic9@mailbox.org \
    --cc=clang-built-linux@googlegroups.com \
    --cc=keescook@chromium.org \
    --cc=linux-kbuild@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=masahiroy@kernel.org \
    --cc=nathan@kernel.org \
    --cc=ndesaulniers@google.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).