From: Niklas Cassel <cassel@kernel.org>
To: "Lorenzo Pieralisi" <lpieralisi@kernel.org>,
"Krzysztof Wilczyński" <kw@linux.com>,
"Manivannan Sadhasivam" <manivannan.sadhasivam@linaro.org>,
"Rob Herring" <robh@kernel.org>,
"Bjorn Helgaas" <bhelgaas@google.com>,
"Jingoo Han" <jingoohan1@gmail.com>,
"Heiko Stuebner" <heiko@sntech.de>,
"Kishon Vijay Abraham I" <kishon@kernel.org>
Cc: Damien Le Moal <dlemoal@kernel.org>,
Siddharth Vadapalli <s-vadapalli@ti.com>,
Udit Kumar <u-kumar1@ti.com>,
Vignesh Raghavendra <vigneshr@ti.com>,
Niklas Cassel <cassel@kernel.org>,
linux-pci@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
linux-rockchip@lists.infradead.org
Subject: [PATCH v4 0/7] PCI: endpoint: Allow EPF drivers to configure the size of Resizable BARs
Date: Fri, 31 Jan 2025 19:29:49 +0100 [thread overview]
Message-ID: <20250131182949.465530-9-cassel@kernel.org> (raw)
The PCI endpoint framework currently does not support resizable BARs.
Add a new BAR type BAR_RESIZABLE, so that EPC drivers can support resizable
BARs properly.
For a resizable BAR, we will only allow a single supported size.
This is by design, as we do not need/want the complexity of the host side
resizing our resizable BAR.
In the DWC driver specifically, the DWC driver currently handles resizable
BARs using an ugly hack where a resizable BAR is force set to a fixed size
BAR with 1 MB size if detected. This is bogus, as a resizable BAR can be
configured to sizes other than 1 MB.
With these changes, an EPF driver will be able to call pci_epc_set_bar()
to configure a resizable BAR to an arbitrary size, just like for
BAR_PROGRAMMABLE. Thus, DWC based EPF drivers will no longer be forced to
a bogus 1 MB forced size for resizable BARs.
Tested/verified on a Radxa Rock 5b (rk3588) by:
-Modifying pci-epf-test.c to request BAR sizes that are larger than 1 MB:
-static size_t bar_size[] = { 512, 512, 1024, 16384, 131072, 1048576 };
+static size_t bar_size[] = { SZ_1M, SZ_1M, SZ_2M, SZ_2M, SZ_4M, SZ_4M };
(Make sure to set CONFIG_CMA_ALIGNMENT=10 such that dma_alloc_coherent()
calls are aligned even for allocations larger than 1 MB.)
-Rebooting the host to make sure that the DWC EP driver configures the BARs
correctly after receiving a link down event.
-Modifying EPC features to configure a BAR as 64-bit, to make sure that we
handle 64-bit BARs correctly.
-Modifying the DWC EP driver to set a size larger than 2 GB, to make sure
we handle BAR sizes larger than 2 GB (for 64-bit BARs) correctly.
-Running the consecutive BAR test in pci_endpoint_test.c to make sure that
the address translation works correctly.
Changes since V3:
-Picked up tags.
-Addressed comments from Mani.
Kind regards,
Niklas
Niklas Cassel (7):
PCI: endpoint: Allow EPF drivers to configure the size of Resizable
BARs
PCI: endpoint: Add pci_epc_bar_size_to_rebar_cap()
PCI: dwc: ep: Move dw_pcie_ep_find_ext_capability()
PCI: dwc: endpoint: Allow EPF drivers to configure the size of
Resizable BARs
PCI: keystone: Describe Resizable BARs as Resizable BARs
PCI: keystone: Specify correct alignment requirement
PCI: dw-rockchip: Describe Resizable BARs as Resizable BARs
drivers/pci/controller/dwc/pci-keystone.c | 6 +-
.../pci/controller/dwc/pcie-designware-ep.c | 218 +++++++++++++++---
drivers/pci/controller/dwc/pcie-dw-rockchip.c | 22 +-
drivers/pci/endpoint/pci-epc-core.c | 31 +++
drivers/pci/endpoint/pci-epf-core.c | 4 +
include/linux/pci-epc.h | 5 +
6 files changed, 239 insertions(+), 47 deletions(-)
--
2.48.1
next reply other threads:[~2025-01-31 18:31 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-01-31 18:29 Niklas Cassel [this message]
2025-01-31 18:29 ` [PATCH v4 7/7] PCI: dw-rockchip: Describe Resizable BARs as Resizable BARs Niklas Cassel
2025-02-13 13:33 ` [PATCH v4 0/7] PCI: endpoint: Allow EPF drivers to configure the size of " Niklas Cassel
2025-02-14 17:40 ` Manivannan Sadhasivam
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=20250131182949.465530-9-cassel@kernel.org \
--to=cassel@kernel.org \
--cc=bhelgaas@google.com \
--cc=dlemoal@kernel.org \
--cc=heiko@sntech.de \
--cc=jingoohan1@gmail.com \
--cc=kishon@kernel.org \
--cc=kw@linux.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux-rockchip@lists.infradead.org \
--cc=lpieralisi@kernel.org \
--cc=manivannan.sadhasivam@linaro.org \
--cc=robh@kernel.org \
--cc=s-vadapalli@ti.com \
--cc=u-kumar1@ti.com \
--cc=vigneshr@ti.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).