All of lore.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Fangrui Song <i@maskray.me>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>,
	Guo Ren <guoren@kernel.org>, Palmer Dabbelt <palmer@dabbelt.com>,
	Charlie Jenkins <thecharlesjenkins@gmail.com>,
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org,
	llvm@lists.linux.dev, Nathan Chancellor <nathan@kernel.org>
Subject: Re: [PATCH] riscv/runtime-const: Disable linker relaxation for RUNTIME_MAGIC
Date: Mon, 3 Aug 2026 10:27:32 +0200	[thread overview]
Message-ID: <20260803082732.GZ49951@noisy.programming.kicks-ass.net> (raw)
In-Reply-To: <20260802174342.597092-1-i@maskray.me>

On Sun, Aug 02, 2026 at 10:43:42AM -0700, Fangrui Song wrote:
> Commit ee10b1028129 ("riscv/runtime-const: Replace open-coded
> placeholder with RUNTIME_MAGIC") switched the lui + addi[w] placeholder
> from hand-encoded immediates to %hi()/%lo() of RUNTIME_MAGIC.
> 
> GNU Assembler folds %hi()/%lo() of an absolute expression at assembly
> time. LLVM's integrated assembler since 21.1 defers it to the linker
> instead, which range checks it: on riscv64 R_RISCV_HI20 must fit a
> signed 20-bit field, and RUNTIME_MAGIC (0x89ABCDEF) is a positive 64-bit
> value rather than a sign-extended 32-bit one, so it does not:
> 
>     ld.lld: error: relocation R_RISCV_HI20 out of range: 563901 is not in
>     [-524288, 524287]
> 
> The sequence is patched at runtime and its instruction offsets are
> recorded via ".long 1b - .", so the linker must not touch it in the
> first place. Add ".option norelax", as is already done for
> ALTERNATIVE() and static keys; the integrated assembler then resolves
> %hi()/%lo() itself and emits no relocation, restoring the exact
> encoding the open-coded placeholder produced.
> 
> Fixes: ee10b1028129 ("riscv/runtime-const: Replace open-coded placeholder with RUNTIME_MAGIC")
> Reported-by: Nathan Chancellor <nathan@kernel.org>
> Closes: https://github.com/ClangBuiltLinux/linux/issues/2179
> Signed-off-by: Fangrui Song <i@maskray.me>
> ---
> Applies to tip/locking/core; the RISC-V tree (riscv/linux.git) does not
> carry ee10b1028129.

Thanks, I'll stick this on top.

WARNING: multiple messages have this Message-ID (diff)
From: Peter Zijlstra <peterz@infradead.org>
To: Fangrui Song <i@maskray.me>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>,
	Guo Ren <guoren@kernel.org>, Palmer Dabbelt <palmer@dabbelt.com>,
	Charlie Jenkins <thecharlesjenkins@gmail.com>,
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org,
	llvm@lists.linux.dev, Nathan Chancellor <nathan@kernel.org>
Subject: Re: [PATCH] riscv/runtime-const: Disable linker relaxation for RUNTIME_MAGIC
Date: Mon, 3 Aug 2026 10:27:32 +0200	[thread overview]
Message-ID: <20260803082732.GZ49951@noisy.programming.kicks-ass.net> (raw)
In-Reply-To: <20260802174342.597092-1-i@maskray.me>

On Sun, Aug 02, 2026 at 10:43:42AM -0700, Fangrui Song wrote:
> Commit ee10b1028129 ("riscv/runtime-const: Replace open-coded
> placeholder with RUNTIME_MAGIC") switched the lui + addi[w] placeholder
> from hand-encoded immediates to %hi()/%lo() of RUNTIME_MAGIC.
> 
> GNU Assembler folds %hi()/%lo() of an absolute expression at assembly
> time. LLVM's integrated assembler since 21.1 defers it to the linker
> instead, which range checks it: on riscv64 R_RISCV_HI20 must fit a
> signed 20-bit field, and RUNTIME_MAGIC (0x89ABCDEF) is a positive 64-bit
> value rather than a sign-extended 32-bit one, so it does not:
> 
>     ld.lld: error: relocation R_RISCV_HI20 out of range: 563901 is not in
>     [-524288, 524287]
> 
> The sequence is patched at runtime and its instruction offsets are
> recorded via ".long 1b - .", so the linker must not touch it in the
> first place. Add ".option norelax", as is already done for
> ALTERNATIVE() and static keys; the integrated assembler then resolves
> %hi()/%lo() itself and emits no relocation, restoring the exact
> encoding the open-coded placeholder produced.
> 
> Fixes: ee10b1028129 ("riscv/runtime-const: Replace open-coded placeholder with RUNTIME_MAGIC")
> Reported-by: Nathan Chancellor <nathan@kernel.org>
> Closes: https://github.com/ClangBuiltLinux/linux/issues/2179
> Signed-off-by: Fangrui Song <i@maskray.me>
> ---
> Applies to tip/locking/core; the RISC-V tree (riscv/linux.git) does not
> carry ee10b1028129.

Thanks, I'll stick this on top.

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

  reply	other threads:[~2026-08-03  8:27 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-08-02 17:43 [PATCH] riscv/runtime-const: Disable linker relaxation for RUNTIME_MAGIC Fangrui Song
2026-08-02 17:43 ` Fangrui Song
2026-08-03  8:27 ` Peter Zijlstra [this message]
2026-08-03  8:27   ` Peter Zijlstra
2026-08-03 16:12 ` Nathan Chancellor
2026-08-03 16:12   ` Nathan Chancellor
2026-08-04 10:31 ` [tip: locking/core] " tip-bot2 for Fangrui Song

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=20260803082732.GZ49951@noisy.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=guoren@kernel.org \
    --cc=i@maskray.me \
    --cc=kprateek.nayak@amd.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-riscv@lists.infradead.org \
    --cc=llvm@lists.linux.dev \
    --cc=nathan@kernel.org \
    --cc=palmer@dabbelt.com \
    --cc=thecharlesjenkins@gmail.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.