From: Mika Westerberg <mika.westerberg@linux.intel.com>
To: Bjorn Helgaas <bhelgaas@google.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>,
Mika Westerberg <mika.westerberg@linux.intel.com>,
linux-pci@vger.kernel.org
Subject: [PATCH] PCI: Quirk Intel DG2 ASPM L1 acceptable latency to be unlimited
Date: Tue, 5 Apr 2022 12:38:10 +0300 [thread overview]
Message-ID: <20220405093810.76613-1-mika.westerberg@linux.intel.com> (raw)
Intel DG2 discrete graphics PCIe endpoints hard-code their acceptable L1
ASPM latency to be < 1us even though the hardware actually supports
higher latencies (> 64 us) just fine. In order to allow the links to go
into L1 and save power, quirk the acceptable L1 ASPM latency for these
endpoints to be unlimited.
Note this does not have any effect unless the user requested the kernel
to enable ASPM in the first place (by default we don't enable it). This
is done with "pcie_aspm=force pcie_aspm.policy=powersupsersave" command
line parameters.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
---
drivers/pci/quirks.c | 44 ++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 44 insertions(+)
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index da829274fc66..e97b5daa00eb 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -5895,3 +5895,47 @@ DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x1533, rom_bar_overlap_defect);
DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x1536, rom_bar_overlap_defect);
DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x1537, rom_bar_overlap_defect);
DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x1538, rom_bar_overlap_defect);
+
+#ifdef CONFIG_PCIEASPM
+/*
+ * Intel DG2 graphics card has hard-coded acceptable L1 latency that is
+ * too low which prevents ASPM to be enabled. It does support ASPM L1
+ * and tolerates higher latencies so quirk it to be unlimited.
+ */
+static void quirk_aspm_accepted_l1_latency(struct pci_dev *dev)
+{
+ if ((dev->devcap & PCI_EXP_DEVCAP_L1) >> 9 < 7) {
+ u32 devcap = dev->devcap;
+
+ dev->devcap |= 7 << 9;
+ pci_info(dev, "quirking devcap for L1 accepted latency 0x%08x -> 0x%08x\n",
+ devcap, dev->devcap);
+ }
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f80, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f81, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f82, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f83, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f84, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f85, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f86, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f87, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x4f88, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x5690, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x5691, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x5692, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x5693, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x5694, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x5695, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a0, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a1, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a2, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a3, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a4, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a5, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56a6, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56b0, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56b1, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56c0, quirk_aspm_accepted_l1_latency);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x56c1, quirk_aspm_accepted_l1_latency);
+#endif
--
2.35.1
next reply other threads:[~2022-04-05 11:08 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-05 9:38 Mika Westerberg [this message]
2022-04-05 12:49 ` [PATCH] PCI: Quirk Intel DG2 ASPM L1 acceptable latency to be unlimited Rodrigo Vivi
2022-04-05 16:01 ` Bjorn Helgaas
2022-04-05 16:23 ` Vivi, Rodrigo
2022-04-05 16:24 ` Mika Westerberg
2022-04-07 17:06 ` Bjorn Helgaas
2022-04-08 6:44 ` Mika Westerberg
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=20220405093810.76613-1-mika.westerberg@linux.intel.com \
--to=mika.westerberg@linux.intel.com \
--cc=bhelgaas@google.com \
--cc=linux-pci@vger.kernel.org \
--cc=rodrigo.vivi@intel.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