From: Julien Olivain via buildroot <buildroot@buildroot.org>
To: Fengwei Tan <tfx2001@outlook.com>
Cc: buildroot@buildroot.org,
Francois Dugast <francois.dugast.foss@gmail.com>
Subject: Re: [Buildroot] [PATCH] configs/visionfive2_defconfig: bump Linux to 6.18.52
Date: Fri, 09 Oct 2026 22:21:11 +0200 [thread overview]
Message-ID: <ea0daeeffd12e78135a8bcc5363fd5e9@free.fr> (raw)
In-Reply-To: <SY8P300MB00411684A8A44ED2157BB57AB5942@SY8P300MB0041.AUSP300.PROD.OUTLOOK.COM>
Hi Fengwei,
On 07/10/2026 19:45, Fengwei Tan wrote:
> When running iperf3 with Linux 6.18.51 on VisionFive 2, one CPU core
> stays at 100% load during the test. The iperf3 server log shows the
> throughput is limited to approximately 793 Mbits/sec:
>
> [ ID] Interval Transfer Bitrate
> [ 5] 0.00-10.01 sec 945 MBytes 793 Mbits/sec
> receiver
>
> The regression was introduced in Linux 6.14 by commit df542f669307
> ("net: stmmac: Switch to zero-copy in non-XDP RX path"). It leads to
> misaligned access exceptions when parsing packet headers. During the
> iperf3 test, perf stat reported approximately 9.6 million alignment
> faults over 5 seconds:
>
> $ perf stat -a -e alignment-faults -- sleep 5
>
> Performance counter stats for 'system wide':
>
> 9620073 alignment-faults
>
> 5.029337000 seconds time elapsed
>
> The kernel must emulate these unaligned accesses, resulting in
> significant overhead.
>
> The Linux kernel CVE team assigned CVE-2026-90059 to this issue [1].
>
> Bump kernel to 6.18.52, which contains the fix for this issue.
>
> [1] https://lore.kernel.org/all/2026091704-CVE-2026-90059-36e6@gregkh
>
> Fixes: 98aa14a186e1 ("configs/visionfive2: bump Linux to 6.18.51,
> U-Boot to 2026.07, and OpenSBI to 1.8.1")
> Assisted-by: Codex:GPT-6.1-Sol
> Signed-off-by: Fengwei Tan <tfx2001@outlook.com>
> ---
> Hello all,
>
> This patch fixes a performance issue on the VisionFive 2, caused by
> misaligned access when the Linux kernel network stack parses packet
> headers.
>
> I noticed this issue while running iperf3 tests on the VisionFive 2
> board. I captured perf data [1] and analyzed it with LLM, which
> quickly identified the relevant upstream kernel commit and the
> corresponding lore archive [2].
>
> I rebuilt Linux 6.18.51 with the fix patch [3] applied and retested;
> the
> issue was resolved. So I am pretty sure this issue was introduced by
> the
> commit identified by the LLM.
>
> [1] https://gist.github.com/tfx2001/f1f10ee1274acb7063ef1def5f2b119a
> [2] https://lore.kernel.org/all/2026091704-CVE-2026-90059-36e6@gregkh/
> [3]
> https://git.kernel.org/stable/c/2ea42936416b29cca42b2c7bc7d3d3768f77c199
> ---
> board/visionfive2/patches/linux/linux.hash | 2 +-
> configs/visionfive2_defconfig | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/board/visionfive2/patches/linux/linux.hash
> b/board/visionfive2/patches/linux/linux.hash
> index b2cfc9b7e7..1e58cd0b9f 100644
> --- a/board/visionfive2/patches/linux/linux.hash
> +++ b/board/visionfive2/patches/linux/linux.hash
> @@ -1,2 +1,2 @@
> # Locally calculated
> -sha256
> ba2f60f858bf4d1f929101faa356c93dc8b925b17aaa9f95eabd4627758df613
> linux-6.18.51.tar.xz
> +sha256
> 2b69564f7d4fea0c859b1959ba33709ee6e9139bd100e30a853b57159a8221b8
> linux-6.18.52.tar.xz
> diff --git a/configs/visionfive2_defconfig
> b/configs/visionfive2_defconfig
> index 7ce605b84b..bfe5e42cc5 100644
> --- a/configs/visionfive2_defconfig
> +++ b/configs/visionfive2_defconfig
> @@ -7,7 +7,7 @@ BR2_ROOTFS_OVERLAY="board/visionfive2/overlay"
> BR2_ROOTFS_POST_IMAGE_SCRIPT="board/visionfive2/post-image.sh"
> BR2_LINUX_KERNEL=y
> BR2_LINUX_KERNEL_CUSTOM_VERSION=y
> -BR2_LINUX_KERNEL_CUSTOM_VERSION_VALUE="6.18.51"
> +BR2_LINUX_KERNEL_CUSTOM_VERSION_VALUE="6.18.52"
Your motivation for this bump is well explained.
Is there any reason to bump to 6.18.52 while there is
the version 6.18.55 already available (which probably
contains many other fixes)?
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.18.55
If there is no specific reason, could you send an updated
patch which bumps to the latest 6.18.y series, please?
> BR2_LINUX_KERNEL_USE_CUSTOM_CONFIG=y
>
> BR2_LINUX_KERNEL_CUSTOM_CONFIG_FILE="board/visionfive2/linux_defconfig"
> BR2_LINUX_KERNEL_DTS_SUPPORT=y
>
> ---
> base-commit: 1ab8f8ef5af5a170a1ecbea3209ca39afd49194c
> change-id: 20261007-visionfive2-bump-linux-version-0de383d488d3
>
> Best regards,
> --
> Fengwei Tan <tfx2001@outlook.com>
Best regards,
Julien.
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot
next prev parent reply other threads:[~2026-10-09 20:21 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-07 17:45 [Buildroot] [PATCH] configs/visionfive2_defconfig: bump Linux to 6.18.52 Fengwei Tan
2026-10-09 20:21 ` Julien Olivain via buildroot [this message]
2026-10-10 3:08 ` Fengwei Tan
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=ea0daeeffd12e78135a8bcc5363fd5e9@free.fr \
--to=buildroot@buildroot.org \
--cc=francois.dugast.foss@gmail.com \
--cc=ju.o@free.fr \
--cc=tfx2001@outlook.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.