From: Nathan Chancellor <nathan@kernel.org>
To: Christoph Hellwig <hch@lst.de>
Cc: iommu@lists.linux-foundation.org, x86@kernel.org,
Anshuman Khandual <anshuman.khandual@arm.com>,
Tom Lendacky <thomas.lendacky@amd.com>,
Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>,
Stefano Stabellini <sstabellini@kernel.org>,
Boris Ostrovsky <boris.ostrovsky@oracle.com>,
Juergen Gross <jgross@suse.com>, Joerg Roedel <joro@8bytes.org>,
David Woodhouse <dwmw2@infradead.org>,
Lu Baolu <baolu.lu@linux.intel.com>,
Robin Murphy <robin.murphy@arm.com>,
linux-arm-kernel@lists.infradead.org,
xen-devel@lists.xenproject.org, linux-ia64@vger.kernel.org,
linux-mips@vger.kernel.org, linuxppc-dev@lists.ozlabs.org,
linux-riscv@lists.infradead.org, linux-s390@vger.kernel.org,
linux-hyperv@vger.kernel.org, tboot-devel@lists.sourceforge.net,
linux-pci@vger.kernel.org
Subject: Re: [PATCH 09/15] swiotlb: make the swiotlb_init interface more useful
Date: Wed, 1 Jun 2022 10:25:31 -0700 [thread overview]
Message-ID: <YpehC7BwBlnuxplF@dev-arch.thelio-3990X> (raw)
In-Reply-To: <20220404050559.132378-10-hch@lst.de>
Hi Christoph,
On Mon, Apr 04, 2022 at 07:05:53AM +0200, Christoph Hellwig wrote:
> Pass a bool to pass if swiotlb needs to be enabled based on the
> addressing needs and replace the verbose argument with a set of
> flags, including one to force enable bounce buffering.
>
> Note that this patch removes the possibility to force xen-swiotlb
> use using swiotlb=force on the command line on x86 (arm and arm64
> never supported that), but this interface will be restored shortly.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
I bisected a performance regression in WSL2 to this change as commit
c6af2aa9ffc9 ("swiotlb: make the swiotlb_init interface more useful") in
mainline (bisect log below). I initially noticed it because accessing the
Windows filesystem through the /mnt/c mount is about 40x slower if I am doing
my math right based on the benchmarks below.
Before:
$ uname -r; and hyperfine "ls -l /mnt/c/Users/natec/Downloads"
5.18.0-rc3-microsoft-standard-WSL2-00008-ga3e230926708
Benchmark 1: ls -l /mnt/c/Users/natec/Downloads
Time (mean ± σ): 564.5 ms ± 24.1 ms [User: 2.5 ms, System: 130.3 ms]
Range (min … max): 510.2 ms … 588.0 ms 10 runs
After
$ uname -r; and hyperfine "ls -l /mnt/c/Users/natec/Downloads"
5.18.0-rc3-microsoft-standard-WSL2-00009-gc6af2aa9ffc9
Benchmark 1: ls -l /mnt/c/Users/natec/Downloads
Time (mean ± σ): 23.282 s ± 1.220 s [User: 0.013 s, System: 0.101 s]
Range (min … max): 21.793 s … 25.317 s 10 runs
I do see 'swiotlb=force' on the cmdline:
$ cat /proc/cmdline
initrd=\initrd.img panic=-1 nr_cpus=8 swiotlb=force earlycon=uart8250,io,0x3f8,115200 console=hvc0 debug pty.legacy_count=0
/mnt/c appears to be a 9p mount, not sure if that is relevant here:
$ mount &| grep /mnt/c
drvfs on /mnt/c type 9p (rw,noatime,dirsync,aname=drvfs;path=C:\;uid=1000;gid=1000;symlinkroot=/mnt/,mmap,access=client,msize=262144,trans=virtio)
If there is any other information I can provide, please let me know.
Cheers,
Nathan
# bad: [700170bf6b4d773e328fa54ebb70ba444007c702] Merge tag 'nfs-for-5.19-1' of git://git.linux-nfs.org/projects/anna/linux-nfs
# good: [4b0986a3613c92f4ec1bdc7f60ec66fea135991f] Linux 5.18
git bisect start '700170bf6b4d773e328fa54ebb70ba444007c702' 'v5.18'
# good: [86c87bea6b42100c67418af690919c44de6ede6e] Merge tag 'devicetree-for-5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux
git bisect good 86c87bea6b42100c67418af690919c44de6ede6e
# bad: [ae862183285cbb2ef9032770d98ffa9becffe9d5] Merge tag 'arm-dt-5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
git bisect bad ae862183285cbb2ef9032770d98ffa9becffe9d5
# good: [2518f226c60d8e04d18ba4295500a5b0b8ac7659] Merge tag 'drm-next-2022-05-25' of git://anongit.freedesktop.org/drm/drm
git bisect good 2518f226c60d8e04d18ba4295500a5b0b8ac7659
# bad: [babf0bb978e3c9fce6c4eba6b744c8754fd43d8e] Merge tag 'xfs-5.19-for-linus' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
git bisect bad babf0bb978e3c9fce6c4eba6b744c8754fd43d8e
# good: [beed983621fbdfd291e6e3a0cdc4d10517e60af8] ASoC: Intel: avs: Machine board registration
git bisect good beed983621fbdfd291e6e3a0cdc4d10517e60af8
# good: [fbe86daca0ba878b04fa241b85e26e54d17d4229] Merge tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
git bisect good fbe86daca0ba878b04fa241b85e26e54d17d4229
# good: [166afc45ed5523298541fd0297f9ad585cc2708c] Merge tag 'reflink-speedups-5.19_2022-04-28' of git://git.kernel.org/pub/scm/linux/kernel/git/djwong/xfs-linux into xfs-5.19-for-next
git bisect good 166afc45ed5523298541fd0297f9ad585cc2708c
# bad: [e375780b631a5fc2a61a3b4fa12429255361a31e] Merge tag 'fsnotify_for_v5.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
git bisect bad e375780b631a5fc2a61a3b4fa12429255361a31e
# bad: [4a37f3dd9a83186cb88d44808ab35b78375082c9] dma-direct: don't over-decrypt memory
git bisect bad 4a37f3dd9a83186cb88d44808ab35b78375082c9
# bad: [742519538e6b07250c8085bbff4bd358bc03bf16] swiotlb: pass a gfp_mask argument to swiotlb_init_late
git bisect bad 742519538e6b07250c8085bbff4bd358bc03bf16
# good: [9bbe7a7fc126e3d14fefa4b035854aba080926d9] arm/xen: don't check for xen_initial_domain() in xen_create_contiguous_region
git bisect good 9bbe7a7fc126e3d14fefa4b035854aba080926d9
# good: [a3e230926708125205ffd06d3dc2175a8263ae7e] x86: centralize setting SWIOTLB_FORCE when guest memory encryption is enabled
git bisect good a3e230926708125205ffd06d3dc2175a8263ae7e
# bad: [8ba2ed1be90fc210126f68186564707478552c95] swiotlb: add a SWIOTLB_ANY flag to lift the low memory restriction
git bisect bad 8ba2ed1be90fc210126f68186564707478552c95
# bad: [c6af2aa9ffc9763826607bc2664ef3ea4475ed18] swiotlb: make the swiotlb_init interface more useful
git bisect bad c6af2aa9ffc9763826607bc2664ef3ea4475ed18
# first bad commit: [c6af2aa9ffc9763826607bc2664ef3ea4475ed18] swiotlb: make the swiotlb_init interface more useful
next prev parent reply other threads:[~2022-06-01 17:25 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-04 5:05 cleanup swiotlb initialization v8 Christoph Hellwig
2022-04-04 5:05 ` [PATCH 01/15] dma-direct: use is_swiotlb_active in dma_direct_map_page Christoph Hellwig
2022-04-04 5:05 ` [PATCH 02/15] swiotlb: make swiotlb_exit a no-op if SWIOTLB_FORCE is set Christoph Hellwig
2022-04-04 5:05 ` [PATCH 03/15] swiotlb: simplify swiotlb_max_segment Christoph Hellwig
2022-04-04 5:05 ` [PATCH 04/15] swiotlb: rename swiotlb_late_init_with_default_size Christoph Hellwig
2022-04-04 5:05 ` [PATCH 05/15] arm/xen: don't check for xen_initial_domain() in xen_create_contiguous_region Christoph Hellwig
2022-04-04 5:05 ` [PATCH 06/15] MIPS/octeon: use swiotlb_init instead of open coding it Christoph Hellwig
2022-04-04 5:05 ` [PATCH 07/15] x86: remove the IOMMU table infrastructure Christoph Hellwig
2022-04-07 0:20 ` Konrad Rzeszutek Wilk
2022-04-04 5:05 ` [PATCH 08/15] x86: centralize setting SWIOTLB_FORCE when guest memory encryption is enabled Christoph Hellwig
2022-04-04 5:05 ` [PATCH 09/15] swiotlb: make the swiotlb_init interface more useful Christoph Hellwig
2022-04-04 17:00 ` Michael Kelley (LINUX)
2022-06-01 17:25 ` Nathan Chancellor [this message]
2022-06-01 17:34 ` Christoph Hellwig
[not found] ` <YpemDuzdoaO3rijX@Ryzen-9-3900X.>
2022-06-01 17:57 ` Christoph Hellwig
2022-06-01 18:11 ` Nathan Chancellor
2022-06-01 18:21 ` Christoph Hellwig
2022-06-01 18:40 ` Nathan Chancellor
2022-04-04 5:05 ` [PATCH 10/15] swiotlb: add a SWIOTLB_ANY flag to lift the low memory restriction Christoph Hellwig
2022-04-07 0:25 ` Konrad Rzeszutek Wilk
2022-04-12 6:21 ` Christoph Hellwig
2022-04-04 5:05 ` [PATCH 11/15] swiotlb: pass a gfp_mask argument to swiotlb_init_late Christoph Hellwig
2022-04-04 5:05 ` [PATCH 12/15] swiotlb: provide swiotlb_init variants that remap the buffer Christoph Hellwig
2022-04-04 7:09 ` Dongli Zhang
2022-04-04 5:05 ` [PATCH 13/15] swiotlb: merge swiotlb-xen initialization into swiotlb Christoph Hellwig
2022-04-04 5:05 ` [PATCH 14/15] swiotlb: remove swiotlb_init_with_tbl and swiotlb_init_late_with_tbl Christoph Hellwig
2022-04-04 5:05 ` [PATCH 15/15] x86: remove cruft from <asm/dma-mapping.h> Christoph Hellwig
[not found] ` <67c1784af6f24f3e871ddfb1478e821c@FR3P281MB0843.DEUP281.PROD.OUTLOOK.COM>
2022-04-04 6:19 ` [PATCH 12/15] swiotlb: provide swiotlb_init variants that remap the buffer Alan Robinson
2022-04-05 20:06 ` cleanup swiotlb initialization v8 Boris Ostrovsky
2022-04-07 0:31 ` Konrad Rzeszutek Wilk
2022-04-13 6:20 ` Christoph Hellwig
-- strict thread matches above, loose matches on Subject: below --
2022-03-14 7:31 cleanup swiotlb initialization v5 Christoph Hellwig
2022-03-14 7:31 ` [PATCH 09/15] swiotlb: make the swiotlb_init interface more useful Christoph Hellwig
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=YpehC7BwBlnuxplF@dev-arch.thelio-3990X \
--to=nathan@kernel.org \
--cc=anshuman.khandual@arm.com \
--cc=baolu.lu@linux.intel.com \
--cc=boris.ostrovsky@oracle.com \
--cc=dwmw2@infradead.org \
--cc=hch@lst.de \
--cc=iommu@lists.linux-foundation.org \
--cc=jgross@suse.com \
--cc=joro@8bytes.org \
--cc=konrad.wilk@oracle.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-hyperv@vger.kernel.org \
--cc=linux-ia64@vger.kernel.org \
--cc=linux-mips@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux-riscv@lists.infradead.org \
--cc=linux-s390@vger.kernel.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=robin.murphy@arm.com \
--cc=sstabellini@kernel.org \
--cc=tboot-devel@lists.sourceforge.net \
--cc=thomas.lendacky@amd.com \
--cc=x86@kernel.org \
--cc=xen-devel@lists.xenproject.org \
/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).