From: Agustin Santiago Isasmendi <isasmendi.agus@gmail.com>
To: Navid Ghahremani <ghahramani.navid@gmail.com>
Cc: stefandoesinger@gmail.com, linux-arm-kernel@lists.infradead.org,
arnd@arndb.de, krzk@kernel.org
Subject: Re: [PATCH v2 00/24] ARM: zte: Add support for Sanechips ZX279128S and ZTE ZXHN H3600
Date: Thu, 19 Nov 2026 11:37:38 +0100 [thread overview]
Message-ID: <20261119103738.3525992-1-isasmendi.agus@gmail.com> (raw)
In-Reply-To: <20261113003635.3664249-1-ghahramani.navid@gmail.com>
Hi Navid,
Thanks for the tests and the numbers.
On Fri, Nov 13, 2026 at 00:36, Navid Ghahremani wrote:
> With the lock on both sides, PCIe config reads of the MT7915 still hang
> my unit.
That fits: a config read is a PCIe read like any other. I hadn't
covered them in my lock either; when they were left unprotected in one
of my tests the board hung the same way.
> Because of these numbers OpenWrt will keep the cache on with the lock.
Makes sense. Two things that may help there.
The stock kernel makes the PCIe DMA hardware-coherent: in
zx279128_init_machine it writes
0x00a20080 = 0x40000001, 0x00a20084 = 0xbfffffff,
0x00a20078 = 0x00a20088 = 0x1f0f1f0f, 0x00a20000 = 0x0d000000
and mt7915.ko uses the coherent DMA ops. With those writes and
dma-coherent on the PCIe nodes the mt7915 needs no cache maintenance
(md5-checked transfers, L2 on and off); without them its firmware does
not load. The switch DMA stays non-coherent. The same setting also
makes the IDM ring coherent (stock uses it through the cacheable alias
with no maintenance), so the WiFi offload RFC has to treat the ring as
coherent when this is on.
On my unit, with the L2 on and the lock (Mbit/s):
plain coherent coherent + offload
TCP router->phone 432 525 530
TCP phone->router 235 283 288
routed WAN->WiFi 253 237 270 (CPU 46% -> 16%)
And the lock: I reworked ours so that only a small hook is left in
mt76 and the rest lives in platform code, which may suit your OpenWrt
PR. The L2C side wraps outer_cache in mach-zte (cache-l2x0.c
untouched), config accesses take the lock in the PCIe host driver
(your lspci case), and mt76 only calls a small outer_io_lock() pair
around its register reads. Your config-read test with WiFi and
Ethernet ran 3 x 10 minutes without a hang, with the same throughput as
the old lock. I can send it to you.
Agustin
prev parent reply other threads:[~2026-11-19 10:37 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-14 7:20 [PATCH v2 00/24] ARM: zte: Add support for Sanechips ZX279128S and ZTE ZXHN H3600 Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 01/24] MAINTAINERS: Consolidate ARM/ZTE ZX SOC SUPPORT with Stefan Doesinger Navid Ghahremani
2026-10-14 20:11 ` Stefan Dösinger
2026-11-13 0:35 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 02/24] dt-bindings: arm: Add ZTE ZX279128S platform descriptions Navid Ghahremani
2026-10-27 21:30 ` Rob Herring (Arm)
2026-10-14 7:20 ` [PATCH v2 03/24] ARM: zte: Add ZX279128S platform and CPU hotplug support Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 04/24] dt-bindings: clock: Add ZTE ZX279128S CRM clock bindings Navid Ghahremani
2026-10-14 21:48 ` Stefan Dösinger
2026-11-13 0:35 ` Navid Ghahremani
2026-10-15 9:25 ` Krzysztof Kozlowski
2026-11-13 0:37 ` Navid Ghahremani
2026-10-15 9:26 ` Krzysztof Kozlowski
2026-10-14 7:20 ` [PATCH v2 05/24] dt-bindings: reset: Add ZTE ZX279128S reset controller bindings Navid Ghahremani
2026-10-15 9:27 ` Krzysztof Kozlowski
2026-11-13 0:37 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 06/24] clk: zte: Add ZX279128S CRM clock and reset driver Navid Ghahremani
2026-10-14 13:21 ` Philipp Zabel
2026-11-13 0:38 ` Navid Ghahremani
2026-10-14 19:39 ` Stefan Dösinger
2026-11-13 0:35 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 07/24] dt-bindings: gpio: Add ZTE ZX279128S GPIO controller Navid Ghahremani
2026-10-27 21:31 ` Rob Herring (Arm)
2026-10-14 7:20 ` [PATCH v2 08/24] gpio: Add ZTE ZX279128S GPIO driver Navid Ghahremani
2026-10-14 21:11 ` Stefan Dösinger
2026-11-13 0:36 ` Navid Ghahremani
2026-10-15 12:25 ` Bartosz Golaszewski
2026-11-13 0:38 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 09/24] dt-bindings: pinctrl: pinctrl-single: Add ZTE ZX279128S pin mux Navid Ghahremani
2026-10-17 16:20 ` Krzysztof Kozlowski
2026-10-14 7:20 ` [PATCH v2 10/24] dt-bindings: phy: Add ZTE ZX279128S PCIe PHY binding Navid Ghahremani
2026-10-15 9:29 ` Krzysztof Kozlowski
2026-11-13 0:37 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 11/24] phy: zte: Add ZTE ZX279128S PCIe PHY driver Navid Ghahremani
2026-11-06 10:13 ` Agustin Santiago Isasmendi
2026-11-13 0:36 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 12/24] dt-bindings: PCI: Add ZTE ZX279128S host controller Navid Ghahremani
2026-10-27 21:40 ` Rob Herring
2026-11-13 0:37 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 13/24] PCI: dwc: Add ZTE ZX279128S host controller driver Navid Ghahremani
2026-10-14 13:19 ` Bjorn Helgaas
2026-10-14 21:06 ` Stefan Dösinger
2026-11-13 0:38 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 14/24] dt-bindings: net: Add ZTE ZX279128S MDIO controller Navid Ghahremani
2026-10-27 21:34 ` Rob Herring (Arm)
2026-10-14 7:20 ` [PATCH v2 15/24] net: mdio: " Navid Ghahremani
2026-10-14 12:13 ` Andrew Lunn
2026-11-13 0:37 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 16/24] net: phy: Add Sanechips ZX5201 PHY support Navid Ghahremani
2026-10-14 12:37 ` Andrew Lunn
2026-11-13 0:36 ` Navid Ghahremani
2026-11-13 1:07 ` Andrew Lunn
2026-11-13 9:56 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 17/24] dt-bindings: net: Add ZTE ZX279128S Ethernet switch Navid Ghahremani
2026-10-27 21:48 ` Rob Herring
2026-11-13 0:37 ` Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 18/24] net: ethernet: zte: Add ZX279128S Ethernet switch driver Navid Ghahremani
2026-10-14 7:20 ` [PATCH v2 19/24] dt-bindings: spi: Add ZTE ZX family SPI flash controller binding Navid Ghahremani
2026-10-14 20:33 ` Stefan Dösinger
2026-10-15 9:35 ` Krzysztof Kozlowski
2026-11-13 0:37 ` Navid Ghahremani
2026-10-15 9:36 ` Krzysztof Kozlowski
2026-11-13 21:09 ` Mark Brown
2026-10-14 7:21 ` [PATCH v2 20/24] spi: zx: Add ZTE ZX family SPI flash controller driver Navid Ghahremani
2026-10-14 19:28 ` Stefan Dösinger
2026-10-15 9:34 ` Krzysztof Kozlowski
2026-11-13 0:37 ` Navid Ghahremani
2026-10-15 14:29 ` kernel test robot
2026-10-19 12:15 ` Stefan Dösinger
2026-11-13 0:35 ` Navid Ghahremani
2026-10-14 7:21 ` [PATCH v2 21/24] dt-bindings: usb: Add ZTE ZX279128S DWC3 controller Navid Ghahremani
2026-10-14 7:21 ` [PATCH v2 22/24] usb: dwc3: generic-plat: Add ZTE ZX279128S Navid Ghahremani
2026-11-04 0:48 ` Thinh Nguyen
2026-11-13 0:38 ` Navid Ghahremani
2026-10-14 7:21 ` [PATCH v2 23/24] ARM: dts: zte: Add ZX279128S SoC description Navid Ghahremani
2026-10-14 7:21 ` [PATCH v2 24/24] ARM: dts: zte: Add ZTE ZXHN H3600 board Navid Ghahremani
2026-10-14 12:52 ` Andrew Lunn
2026-11-13 0:37 ` Navid Ghahremani
2026-11-06 10:13 ` Agustin Santiago Isasmendi
2026-11-13 0:36 ` Navid Ghahremani
2026-10-14 20:00 ` [PATCH v2 00/24] ARM: zte: Add support for Sanechips ZX279128S and ZTE ZXHN H3600 Stefan Dösinger
2026-11-13 0:35 ` Navid Ghahremani
2026-11-14 12:52 ` Stefan Dösinger
2026-11-06 10:11 ` Agustin Santiago Isasmendi
2026-11-13 0:36 ` Navid Ghahremani
2026-11-06 19:08 ` Agustin Santiago Isasmendi
2026-11-08 11:09 ` Stefan Dösinger
2026-11-10 15:35 ` Agustin Santiago Isasmendi
2026-11-11 22:02 ` Stefan Dösinger
2026-11-12 16:10 ` Agustin Santiago Isasmendi
2026-11-13 0:36 ` Navid Ghahremani
2026-11-19 10:37 ` Agustin Santiago Isasmendi [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=20261119103738.3525992-1-isasmendi.agus@gmail.com \
--to=isasmendi.agus@gmail.com \
--cc=arnd@arndb.de \
--cc=ghahramani.navid@gmail.com \
--cc=krzk@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=stefandoesinger@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox