From: Jim Quinlan <james.quinlan@broadcom.com>
To: linux-pci@vger.kernel.org,
Nicolas Saenz Julienne <nsaenz@kernel.org>,
Bjorn Helgaas <bhelgaas@google.com>,
Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
Cyril Brulebois <kibi@debian.org>,
Phil Elwell <phil@raspberrypi.com>,
bcm-kernel-feedback-list@broadcom.com,
james.quinlan@broadcom.com
Cc: "Florian Fainelli" <florian.fainelli@broadcom.com>,
"Jim Quinlan" <jim2101024@gmail.com>,
"Lorenzo Pieralisi" <lpieralisi@kernel.org>,
"Krzysztof Wilczyński" <kw@linux.com>,
"Rob Herring" <robh@kernel.org>,
linux-rpi-kernel@lists.infradead.org (moderated list:BROADCOM
BCM2711/BCM2835 ARM ARCHITECTURE),
linux-arm-kernel@lists.infradead.org (moderated list:BROADCOM
BCM2711/BCM2835 ARM ARCHITECTURE),
linux-kernel@vger.kernel.org (open list)
Subject: [PATCH v9 3/4] PCI: brcmstb: Set downstream maximum {no-}snoop LTR values
Date: Wed, 3 Apr 2024 17:39:00 -0400 [thread overview]
Message-ID: <20240403213902.26391-4-james.quinlan@broadcom.com> (raw)
In-Reply-To: <20240403213902.26391-1-james.quinlan@broadcom.com>
[-- Attachment #1: Type: text/plain, Size: 3929 bytes --]
Most of our systems do not have FW or ACPI, so it is up to the RC driver to
set the maximum LTR {no-}snoop latency values of downstream devices. We
set them to a value that is slightly smaller than the value of our internal
bus timeout register.
Signed-off-by: Jim Quinlan <james.quinlan@broadcom.com>
---
drivers/pci/controller/pcie-brcmstb.c | 68 ++++++++++++++++++++++++++-
1 file changed, 66 insertions(+), 2 deletions(-)
diff --git a/drivers/pci/controller/pcie-brcmstb.c b/drivers/pci/controller/pcie-brcmstb.c
index e3480ca4cd57..3d08b92d5bb8 100644
--- a/drivers/pci/controller/pcie-brcmstb.c
+++ b/drivers/pci/controller/pcie-brcmstb.c
@@ -182,6 +182,20 @@
#define DATA_ADDR(pcie) (pcie->reg_offsets[EXT_CFG_DATA])
#define PCIE_RGR1_SW_INIT_1(pcie) (pcie->reg_offsets[RGR1_SW_INIT_1])
+/*
+ * What we call "LTR_FMT" is the 16 bit latency field format:
+ * [15:15] Requirement bit
+ * [12:10] Latency scale
+ * [09:00] Latency value
+ */
+#define LTR_FMT_TO_NS(p) (FIELD_GET(GENMASK(15, 15), (p)) \
+ * ((unsigned long long)FIELD_GET(GENMASK(9, 0), (p)) \
+ << (FIELD_GET(GENMASK(12, 10), (p)) * 5)))
+#define BRCM_LTR_MAX_SCALE 4 /* Scale==4 => Each unit is 1,048,576ns */
+#define BRCM_LTR_MAX_VALUE 9 /* Using the above scale, roughly 9.4 ms */
+#define BRCM_LTR_MAX_NS ((unsigned long long)(BRCM_LTR_MAX_VALUE \
+ << (5 * BRCM_LTR_MAX_SCALE)))
+
/* Rescal registers */
#define PCIE_DVT_PMU_PCIE_PHY_CTRL 0xc700
#define PCIE_DVT_PMU_PCIE_PHY_CTRL_DAST_NFLDS 0x3
@@ -679,6 +693,49 @@ static void brcm_extend_internal_bus_timeout(struct brcm_pcie *pcie, u32 nsec)
writel(216 * timeout_us, pcie->base + REG_OFFSET);
}
+/* Sets downstream device latency tolerance registers to max we can handle */
+static int brcm_set_dev_ltr_max(struct pci_dev *dev, void *data)
+{
+ u16 ltr_cap_offset = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_LTR);
+ u16 ltr_fmt_cur, ltr_fmt = *(u16 *)data;
+ unsigned long long cur_nsec;
+ static const u16 reg_offsets[2] = {
+ PCI_LTR_MAX_SNOOP_LAT,
+ PCI_LTR_MAX_NOSNOOP_LAT,
+ };
+ unsigned int i;
+
+ if (!ltr_cap_offset || !dev->ltr_path)
+ return 0;
+
+ /*
+ * FW may have already written a value so we want to respect that
+ * value if it is lower than ltr_fmt. Update the current value if
+ * it is 0 or if the new value is less than the current.
+ */
+ for (i = 0; i < ARRAY_SIZE(reg_offsets); i++) {
+ pci_read_config_word(dev, ltr_cap_offset + reg_offsets[i],
+ <r_fmt_cur);
+ cur_nsec = LTR_FMT_TO_NS(ltr_fmt_cur);
+ if (cur_nsec == 0 || cur_nsec > BRCM_LTR_MAX_NS)
+ pci_write_config_word(dev, ltr_cap_offset
+ + reg_offsets[i], ltr_fmt);
+ }
+
+ return 0;
+}
+
+void brcm_set_downstream_devs_ltr_max(struct brcm_pcie *pcie)
+{
+ struct pci_host_bridge *bridge = pci_host_bridge_from_priv(pcie);
+ u16 ltr_fmt = FIELD_PREP(GENMASK(9, 0), BRCM_LTR_MAX_VALUE)
+ | FIELD_PREP(GENMASK(12, 10), BRCM_LTR_MAX_SCALE)
+ | GENMASK(15, 15);
+
+ if (bridge->native_ltr)
+ pci_walk_bus(bridge->bus, brcm_set_dev_ltr_max, <r_fmt);
+}
+
/* The controller is capable of serving in both RC and EP roles */
static bool brcm_pcie_rc_mode(struct brcm_pcie *pcie)
{
@@ -1074,8 +1131,12 @@ static int brcm_pcie_start_link(struct brcm_pcie *pcie)
return -ENODEV;
}
- /* Extend internal bus timeout to 8ms or so */
- brcm_extend_internal_bus_timeout(pcie, SZ_8M);
+ /*
+ * Extend internal bus timeout to 8-10ms, specifically to a value
+ * that is slightly larger than what we are using for the max
+ * {no-}snoop latency we will set in downstream devices.
+ */
+ brcm_extend_internal_bus_timeout(pcie, BRCM_LTR_MAX_NS + 1000);
if (pcie->gen)
brcm_pcie_set_gen(pcie, pcie->gen);
@@ -1616,6 +1677,9 @@ static int brcm_pcie_probe(struct platform_device *pdev)
return ret;
}
+ if (IS_ENABLED(CONFIG_PCIEASPM))
+ brcm_set_downstream_devs_ltr_max(pcie);
+
return 0;
fail:
--
2.17.1
[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4210 bytes --]
next prev parent reply other threads:[~2024-04-03 21:39 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-03 21:38 [PATCH v9 0/4] PCI: brcmstb: Configure appropriate HW CLKREQ# mode Jim Quinlan
2024-04-03 21:38 ` [PATCH v9 1/4] dt-bindings: PCI: brcmstb: Add property "brcm,clkreq-mode" Jim Quinlan
2024-04-03 21:38 ` [PATCH v9 2/4] PCI: brcmstb: Set reasonable value for internal bus timeout Jim Quinlan
2024-05-06 22:45 ` Bjorn Helgaas
2024-04-03 21:39 ` Jim Quinlan [this message]
2024-04-04 16:47 ` [PATCH v9 3/4] PCI: brcmstb: Set downstream maximum {no-}snoop LTR values kernel test robot
2024-04-04 16:48 ` kernel test robot
2024-04-03 21:39 ` [PATCH v9 4/4] PCI: brcmstb: Configure HW CLKREQ# mode appropriate for downstream device Jim Quinlan
2024-05-06 23:20 ` Bjorn Helgaas
2024-05-08 17:55 ` Jim Quinlan
2024-05-08 23:33 ` Bjorn Helgaas
2024-04-04 20:01 ` [PATCH v9 0/4] PCI: brcmstb: Configure appropriate HW CLKREQ# mode Cyril Brulebois
2024-04-30 21:02 ` Jim Quinlan
2024-05-06 22:31 ` Bjorn Helgaas
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=20240403213902.26391-4-james.quinlan@broadcom.com \
--to=james.quinlan@broadcom.com \
--cc=bcm-kernel-feedback-list@broadcom.com \
--cc=bhelgaas@google.com \
--cc=florian.fainelli@broadcom.com \
--cc=jim2101024@gmail.com \
--cc=kibi@debian.org \
--cc=kw@linux.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux-rpi-kernel@lists.infradead.org \
--cc=lorenzo.pieralisi@arm.com \
--cc=lpieralisi@kernel.org \
--cc=nsaenz@kernel.org \
--cc=phil@raspberrypi.com \
--cc=robh@kernel.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