[PATCH 02/13] PCI: aardvark: Fix PCIe Max Payload Size setting

[PATCH 02/13] PCI: aardvark: Fix PCIe Max Payload Size setting

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

Change PCIe Max Payload Size setting in PCIe Device Control register to 512
bytes to align with PCIe Link Initialization sequence as defined in Marvell
Armada 3700 Functional Specification. According to the specification,
maximal Max Payload Size supported by this device is 512 bytes.

Without this kernel prints suspicious line:

    pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 256 (was 16384, max 512)

With this change it changes to:

    pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 256 (was 512, max 512)

Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: stable@vger.kernel.org
---
 drivers/pci/controller/pci-aardvark.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index 596ebcfcc82d..884510630bae 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -488,8 +488,9 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_DEVCTL);
 	reg &= ~PCI_EXP_DEVCTL_RELAX_EN;
 	reg &= ~PCI_EXP_DEVCTL_NOSNOOP_EN;
+	reg &= ~PCI_EXP_DEVCTL_PAYLOAD;
 	reg &= ~PCI_EXP_DEVCTL_READRQ;
-	reg |= PCI_EXP_DEVCTL_PAYLOAD; /* Set max payload size */
+	reg |= PCI_EXP_DEVCTL_PAYLOAD_512B;
 	reg |= PCI_EXP_DEVCTL_READRQ_512B;
 	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_DEVCTL);
 
-- 
2.32.0

[PATCH 05/13] PCI: aardvark: Fix configuring Reference clock

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

Commit 366697018c9a ("PCI: aardvark: Add PHY support") introduced
configuration of PCIe Reference clock via PCIE_CORE_REF_CLK_REG register,
but did it incorrectly.

PCIe Reference clock differential pair is routed from system board to
endpoint card, so on CPU side it has output direction. Therefore it is
required to enable transmitting and disable receiving.

Default configuration according to Armada 3700 Functional Specifications is
enabled receiver part and disabled transmitter.

We need this change because otherwise PCIe Reference clock is configured to
some undefined state when differential pair is used for both transmitting
and receiving.

Fix this by disabling receiver part.

Fixes: 366697018c9a ("PCI: aardvark: Add PHY support")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: stable@vger.kernel.org
---
 drivers/pci/controller/pci-aardvark.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index bb57ca6aed35..d5d6f92e5143 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -99,6 +99,7 @@
 #define     PCIE_CORE_CTRL2_MSI_ENABLE		BIT(10)
 #define PCIE_CORE_REF_CLK_REG			(CONTROL_BASE_ADDR + 0x14)
 #define     PCIE_CORE_REF_CLK_TX_ENABLE		BIT(1)
+#define     PCIE_CORE_REF_CLK_RX_ENABLE		BIT(2)
 #define PCIE_MSG_LOG_REG			(CONTROL_BASE_ADDR + 0x30)
 #define PCIE_ISR0_REG				(CONTROL_BASE_ADDR + 0x40)
 #define PCIE_MSG_PM_PME_MASK			BIT(7)
@@ -451,9 +452,15 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	u32 reg;
 	int i;
 
-	/* Enable TX */
+	/*
+	 * Configure PCIe Reference clock. Direction is from the PCIe
+	 * controller to the endpoint card, so enable transmitting of
+	 * Reference clock differential signal off-chip and disable
+	 * receiving off-chip differential signal.
+	 */
 	reg = advk_readl(pcie, PCIE_CORE_REF_CLK_REG);
 	reg |= PCIE_CORE_REF_CLK_TX_ENABLE;
+	reg &= ~PCIE_CORE_REF_CLK_RX_ENABLE;
 	advk_writel(pcie, reg, PCIE_CORE_REF_CLK_REG);
 
 	/* Set to Direct mode */
-- 
2.32.0

[PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

It is incorrect to clear status bits of masked interrupts.

The aardvark driver clears all status interrupt bits if no unmasked
status bit is set. Masked bits should never be cleared.

Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: stable@vger.kernel.org
---
 drivers/pci/controller/pci-aardvark.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index d5d6f92e5143..e4986806a189 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -1295,11 +1295,8 @@ static void advk_pcie_handle_int(struct advk_pcie *pcie)
 	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
 	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
 
-	if (!isr0_status && !isr1_status) {
-		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
-		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
+	if (!isr0_status && !isr1_status)
 		return;
-	}
 
 	/* Process MSI interrupts */
 	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
-- 
2.32.0

[PATCH 07/13] PCI: aardvark: Do not unmask unused interrupts

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

There are lot of undocumented interrupt bits. Fix all *_ALL_MASK macros to
define all interrupt bits, so that driver can properly mask all interrupts,
including those which are undocumented.

Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: stable@vger.kernel.org
---
 drivers/pci/controller/pci-aardvark.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index e4986806a189..46fbaba1e6e4 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -107,13 +107,13 @@
 #define     PCIE_ISR0_MSI_INT_PENDING		BIT(24)
 #define     PCIE_ISR0_INTX_ASSERT(val)		BIT(16 + (val))
 #define     PCIE_ISR0_INTX_DEASSERT(val)	BIT(20 + (val))
-#define	    PCIE_ISR0_ALL_MASK			GENMASK(26, 0)
+#define     PCIE_ISR0_ALL_MASK			GENMASK(31, 0)
 #define PCIE_ISR1_REG				(CONTROL_BASE_ADDR + 0x48)
 #define PCIE_ISR1_MASK_REG			(CONTROL_BASE_ADDR + 0x4C)
 #define     PCIE_ISR1_POWER_STATE_CHANGE	BIT(4)
 #define     PCIE_ISR1_FLUSH			BIT(5)
 #define     PCIE_ISR1_INTX_ASSERT(val)		BIT(8 + (val))
-#define     PCIE_ISR1_ALL_MASK			GENMASK(11, 4)
+#define     PCIE_ISR1_ALL_MASK			GENMASK(31, 0)
 #define PCIE_MSI_ADDR_LOW_REG			(CONTROL_BASE_ADDR + 0x50)
 #define PCIE_MSI_ADDR_HIGH_REG			(CONTROL_BASE_ADDR + 0x54)
 #define PCIE_MSI_STATUS_REG			(CONTROL_BASE_ADDR + 0x58)
@@ -199,7 +199,7 @@
 #define     PCIE_IRQ_MSI_INT2_DET		BIT(21)
 #define     PCIE_IRQ_RC_DBELL_DET		BIT(22)
 #define     PCIE_IRQ_EP_STATUS			BIT(23)
-#define     PCIE_IRQ_ALL_MASK			0xfff0fb
+#define     PCIE_IRQ_ALL_MASK			GENMASK(31, 0)
 #define     PCIE_IRQ_ENABLE_INTS_MASK		PCIE_IRQ_CORE_INT
 
 /* Transaction types */
-- 
2.32.0

[PATCH 11/13] PCI: aardvark: Fix link training

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

Fix multiple link training issues in aardvark driver. The main reason of
these issues was misunderstanding of what certain registers do, since their
names and comments were misleading: before commit 96be36dbffac ("PCI:
aardvark: Replace custom macros by standard linux/pci_regs.h macros"), the
pci-aardvark.c driver used custom macros for accessing standard PCIe Root
Bridge registers, and misleading comments did not help to understand what
the code was really doing.

After doing more tests and experiments I've come to the conclusion that the
SPEED_GEN register in aardvark sets the PCIe revision / generation
compliance and forces maximal link speed. Both GEN3 and GEN2 values set the
read-only PCI_EXP_FLAGS_VERS bits (PCIe capabilities version of Root
Bridge) to value 2, while GEN1 value sets PCI_EXP_FLAGS_VERS to 1, which
matches with PCI Express specifications revisions 3, 2 and 1 respectively.
Changing SPEED_GEN also sets the read-only bits PCI_EXP_LNKCAP_SLS and
PCI_EXP_LNKCAP2_SLS to corresponding speed.

(Note that PCI Express rev 1 specification does not define PCI_EXP_LNKCAP2
 and PCI_EXP_LNKCTL2 registers and when SPEED_GEN is set to GEN1 (which
 also sets PCI_EXP_FLAGS_VERS set to 1), lspci cannot access
 PCI_EXP_LNKCAP2 and PCI_EXP_LNKCTL2 registers.)

Changing PCIe link speed can be done via PCI_EXP_LNKCTL2_TLS bits of
PCI_EXP_LNKCTL2 register. Armada 3700 Functional Specifications says that
the default value of PCI_EXP_LNKCTL2_TLS is based on SPEED_GEN value, but
tests showed that the default value is always 8.0 GT/s, independently of
speed set by SPEED_GEN. So after setting SPEED_GEN, we must also set value
in PCI_EXP_LNKCTL2 register via PCI_EXP_LNKCTL2_TLS bits.

Triggering PCI_EXP_LNKCTL_RL bit immediately after setting LINK_TRAINING_EN
bit actually doesn't do anything. Tests have shown that a delay is needed
after enabling LINK_TRAINING_EN bit. As triggering PCI_EXP_LNKCTL_RL
currently does nothing, remove it.

Commit 43fc679ced18 ("PCI: aardvark: Improve link training") introduced
code which sets SPEED_GEN register based on negotiated link speed from
PCI_EXP_LNKSTA_CLS bits of PCI_EXP_LNKSTA register. This code was added to
fix detection of Compex WLE900VX (Atheros QCA9880) WiFi GEN1 PCIe cards, as
otherwise these cards were "invisible" on PCIe bus (probably because they
crashed). But apparently more people reported the same issues with these
cards also with other PCIe controllers [1] and I was able to reproduce this
issue also with other "noname" WiFi cards based on Atheros QCA9890 chip
(with the same PCI vendor/device ids as Atheros QCA9880). So this is not an
issue in aardvark but rather an issue in Atheros QCA98xx chips. Also, this
issue only exists if the kernel is compiled with PCIe ASPM support, and a
generic workaround for this is to change PCIe Bridge to 2.5 GT/s link speed
via PCI_EXP_LNKCTL2_TLS_2_5GT bits in PCI_EXP_LNKCTL2 register [2], before
triggering PCI_EXP_LNKCTL_RL bit. This workaround also works when SPEED_GEN
is set to value GEN2 (5 GT/s). So remove this hack completely in the
aardvark driver and always set SPEED_GEN to value from 'max-link-speed' DT
property. Fix for Atheros QCA98xx chips is handled separately by patch [2].

These two things (code for triggering PCI_EXP_LNKCTL_RL bit and changing
SPEED_GEN value) also explain why commit 6964494582f5 ("PCI: aardvark:
Train link immediately after enabling training") somehow fixed detection of
those problematic Compex cards with Atheros chips: if triggering link
retraining (via PCI_EXP_LNKCTL_RL bit) was done immediately after enabling
link training (via LINK_TRAINING_EN), it did nothing. If there was a
specific delay, aardvark HW already initialized PCIe link and therefore
triggering link retraining caused the above issue. Compex cards triggered
link down event and disappeared from the PCIe bus.

Commit f4c7d053d7f7 ("PCI: aardvark: Wait for endpoint to be ready before
training link") added 100ms sleep before calling 'Start link training'
command and explained that it is a requirement of PCI Express
specification. But the code after this 100ms sleep was not doing 'Start
link training', rather it triggered PCI_EXP_LNKCTL_RL bit via PCIe Root
Bridge to put link into Recovery state.

The required delay after fundamental reset is already done in function
advk_pcie_wait_for_link() which also checks whether PCIe link is up.
So after removing the code which triggers PCI_EXP_LNKCTL_RL bit on PCIe
Root Bridge, there is no need to wait 100ms again. Remove the extra
msleep() call and update comment about the delay required by the PCI
Express specification.

According to Marvell Armada 3700 Functional Specifications, Link training
should be enabled via aardvark register LINK_TRAINING_EN after selecting
PCIe generation and x1 lane. There is no need to disable it prior resetting
card via PERST# signal. This disabling code was introduced in commit
5169a9851daa ("PCI: aardvark: Issue PERST via GPIO") as a workaround for
some Atheros cards. It turns out that this also is Atheros specific issue
and affects any PCIe controller, not only aardvark. Moreover this Atheros
issue was triggered by juggling with PCI_EXP_LNKCTL_RL, LINK_TRAINING_EN
and SPEED_GEN bits interleaved with sleeps. Now, after removing triggering
PCI_EXP_LNKCTL_RL, there is no need to explicitly disable LINK_TRAINING_EN
bit. So remove this code too. The problematic Compex cards described in
previous git commits are correctly detected in advk_pcie_train_link()
function even after applying all these changes.

Note that with this patch, and also prior this patch, some NVMe disks which
support PCIe GEN3 with 8 GT/s speed are negotiated only at the lowest link
speed 2.5 GT/s, independently of SPEED_GEN value. After manually triggering
PCI_EXP_LNKCTL_RL bit (e.g. from userspace via setpci), these NVMe disks
change link speed to 5 GT/s when SPEED_GEN was configured to GEN2. This
issue first needs to be properly investigated. I will send a fix in the
future.

On the other hand, some other GEN2 PCIe cards with 5 GT/s speed are
autonomously by HW autonegotiated at full 5 GT/s speed without need of any
software interaction.

Armada 3700 Functional Specifications describes the following steps for
link training: set SPEED_GEN to GEN2, enable LINK_TRAINING_EN, poll until
link training is complete, trigger PCI_EXP_LNKCTL_RL, poll until signal
rate is 5 GT/s, poll until link training is complete, enable ASPM L0s.

The requirement for triggering PCI_EXP_LNKCTL_RL can be explained by the
need to achieve 5 GT/s speed (as changing link speed is done by throw to
recovery state entered by PCI_EXP_LNKCTL_RL) or maybe as a part of enabling
ASPM L0s (but in this case ASPM L0s should have been enabled prior
PCI_EXP_LNKCTL_RL).

It is unknown why the original pci-aardvark.c driver was triggering
PCI_EXP_LNKCTL_RL bit before waiting for the link to be up. This does not
align with neither PCIe base specifications nor with Armada 3700 Functional
Specification. (Note that in older versions of aardvark, this bit was
called incorrectly PCIE_CORE_LINK_TRAINING, so this may be the reason.)

It is also unknown why Armada 3700 Functional Specification says that it is
needed to trigger PCI_EXP_LNKCTL_RL for GEN2 mode, as according to PCIe
base specification 5 GT/s speed negotiation is supposed to be entirely
autonomous, even if initial speed is 2.5 GT/s.

[1] - https://lore.kernel.org/linux-pci/87h7l8axqp.fsf@toke.dk/
[2] - https://lore.kernel.org/linux-pci/20210326124326.21163-1-pali@kernel.org/

Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: stable@vger.kernel.org # f4c7d053d7f7 ("PCI: aardvark: Wait for endpoint to be ready before training link")
Cc: stable@vger.kernel.org # 6964494582f5 ("PCI: aardvark: Train link immediately after enabling training")
Cc: stable@vger.kernel.org # 43fc679ced18 ("PCI: aardvark: Improve link training")
Cc: stable@vger.kernel.org # 5169a9851daa ("PCI: aardvark: Issue PERST via GPIO")
Cc: stable@vger.kernel.org # 96be36dbffac ("PCI: aardvark: Replace custom macros by standard linux/pci_regs.h macros")
Cc: stable@vger.kernel.org # d0c6a3475b03 ("PCI: aardvark: Move PCIe reset card code to advk_pcie_train_link()")
Cc: stable@vger.kernel.org # 1d1cd163d0de ("PCI: aardvark: Update comment about disabling link training")
---
 drivers/pci/controller/pci-aardvark.c | 117 ++++++++------------------
 1 file changed, 34 insertions(+), 83 deletions(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index 74d1ec7eff16..5387d9cc3eba 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -258,11 +258,6 @@ static inline u32 advk_readl(struct advk_pcie *pcie, u64 reg)
 	return readl(pcie->base + reg);
 }
 
-static inline u16 advk_read16(struct advk_pcie *pcie, u64 reg)
-{
-	return advk_readl(pcie, (reg & ~0x3)) >> ((reg & 0x3) * 8);
-}
-
 static int advk_pcie_link_up(struct advk_pcie *pcie)
 {
 	u32 val, ltssm_state;
@@ -300,23 +295,9 @@ static void advk_pcie_wait_for_retrain(struct advk_pcie *pcie)
 
 static void advk_pcie_issue_perst(struct advk_pcie *pcie)
 {
-	u32 reg;
-
 	if (!pcie->reset_gpio)
 		return;
 
-	/*
-	 * As required by PCI Express spec (PCI Express Base Specification, REV.
-	 * 4.0 PCI Express, February 19 2014, 6.6.1 Conventional Reset) a delay
-	 * for at least 100ms after de-asserting PERST# signal is needed before
-	 * link training is enabled. So ensure that link training is disabled
-	 * prior de-asserting PERST# signal to fulfill that PCI Express spec
-	 * requirement.
-	 */
-	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
-	reg &= ~LINK_TRAINING_EN;
-	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
-
 	/* 10ms delay is needed for some cards */
 	dev_info(&pcie->pdev->dev, "issuing PERST via reset GPIO for 10ms\n");
 	gpiod_set_value_cansleep(pcie->reset_gpio, 1);
@@ -324,53 +305,46 @@ static void advk_pcie_issue_perst(struct advk_pcie *pcie)
 	gpiod_set_value_cansleep(pcie->reset_gpio, 0);
 }
 
-static int advk_pcie_train_at_gen(struct advk_pcie *pcie, int gen)
+static void advk_pcie_train_link(struct advk_pcie *pcie)
 {
-	int ret, neg_gen;
+	struct device *dev = &pcie->pdev->dev;
 	u32 reg;
+	int ret;
 
-	/* Setup link speed */
+	/*
+	 * Setup PCIe rev / gen compliance based on device tree property
+	 * 'max-link-speed' which also forces maximal link speed.
+	 */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
 	reg &= ~PCIE_GEN_SEL_MSK;
-	if (gen == 3)
+	if (pcie->link_gen == 3)
 		reg |= SPEED_GEN_3;
-	else if (gen == 2)
+	else if (pcie->link_gen == 2)
 		reg |= SPEED_GEN_2;
 	else
 		reg |= SPEED_GEN_1;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
 
 	/*
-	 * Enable link training. This is not needed in every call to this
-	 * function, just once suffices, but it does not break anything either.
+	 * Set maximal link speed value also into PCIe Link Control 2 register.
+	 * Armada 3700 Functional Specification says that default value is based
+	 * on SPEED_GEN but tests showed that default value is always 8.0 GT/s.
 	 */
+	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL2);
+	reg &= ~PCI_EXP_LNKCTL2_TLS;
+	if (pcie->link_gen == 3)
+		reg |= PCI_EXP_LNKCTL2_TLS_8_0GT;
+	else if (pcie->link_gen == 2)
+		reg |= PCI_EXP_LNKCTL2_TLS_5_0GT;
+	else
+		reg |= PCI_EXP_LNKCTL2_TLS_2_5GT;
+	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL2);
+
+	/* Enable link training after selecting PCIe generation */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
 	reg |= LINK_TRAINING_EN;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
 
-	/*
-	 * Start link training immediately after enabling it.
-	 * This solves problems for some buggy cards.
-	 */
-	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL);
-	reg |= PCI_EXP_LNKCTL_RL;
-	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL);
-
-	ret = advk_pcie_wait_for_link(pcie);
-	if (ret)
-		return ret;
-
-	reg = advk_read16(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKSTA);
-	neg_gen = reg & PCI_EXP_LNKSTA_CLS;
-
-	return neg_gen;
-}
-
-static void advk_pcie_train_link(struct advk_pcie *pcie)
-{
-	struct device *dev = &pcie->pdev->dev;
-	int neg_gen = -1, gen;
-
 	/*
 	 * Reset PCIe card via PERST# signal. Some cards are not detected
 	 * during link training when they are in some non-initial state.
@@ -381,41 +355,18 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 	 * PERST# signal could have been asserted by pinctrl subsystem before
 	 * probe() callback has been called or issued explicitly by reset gpio
 	 * function advk_pcie_issue_perst(), making the endpoint going into
-	 * fundamental reset. As required by PCI Express spec a delay for at
-	 * least 100ms after such a reset before link training is needed.
-	 */
-	msleep(PCI_PM_D3COLD_WAIT);
-
-	/*
-	 * Try link training at link gen specified by device tree property
-	 * 'max-link-speed'. If this fails, iteratively train at lower gen.
-	 */
-	for (gen = pcie->link_gen; gen > 0; --gen) {
-		neg_gen = advk_pcie_train_at_gen(pcie, gen);
-		if (neg_gen > 0)
-			break;
-	}
-
-	if (neg_gen < 0)
-		goto err;
-
-	/*
-	 * After successful training if negotiated gen is lower than requested,
-	 * train again on negotiated gen. This solves some stability issues for
-	 * some buggy gen1 cards.
+	 * fundamental reset. As required by PCI Express spec (PCI Express
+	 * Base Specification, REV. 4.0 PCI Express, February 19 2014, 6.6.1
+	 * Conventional Reset) a delay for at least 100ms after such a reset
+	 * before sending a Configuration Request to the device is needed.
+	 * So wait until PCIe link is up. Function advk_pcie_wait_for_link()
+	 * waits for link at least 900ms.
 	 */
-	if (neg_gen < gen) {
-		gen = neg_gen;
-		neg_gen = advk_pcie_train_at_gen(pcie, gen);
-	}
-
-	if (neg_gen == gen) {
-		dev_info(dev, "link up at gen %i\n", gen);
-		return;
-	}
-
-err:
-	dev_err(dev, "link never came up\n");
+	ret = advk_pcie_wait_for_link(pcie);
+	if (ret < 0)
+		dev_err(dev, "link never came up\n");
+	else
+		dev_info(dev, "link up\n");
 }
 
 /*
-- 
2.32.0

[PATCH 12/13] PCI: aardvark: Fix checking for link up via LTSSM state

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

Current implementation of advk_pcie_link_up() is wrong as it marks also
link disabled or hot reset states as link up.

Fix it by marking link up only to those states which are defined in PCIe
Base specification 3.0, Table 4-14: Link Status Mapped to the LTSSM.

To simplify implementation, Define macros for every LTSSM state which
aardvark hardware can return in CFG_REG register.

Fix also checking for link training according to the same Table 4-14.
Define a new function advk_pcie_link_training() for this purpose.

Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: Remi Pommarel <repk@triplefau.lt>
Cc: stable@vger.kernel.org
---
 drivers/pci/controller/pci-aardvark.c | 69 +++++++++++++++++++++++++--
 1 file changed, 64 insertions(+), 5 deletions(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index 5387d9cc3eba..9465b630cede 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -165,7 +165,44 @@
 #define CFG_REG					(LMI_BASE_ADDR + 0x0)
 #define     LTSSM_SHIFT				24
 #define     LTSSM_MASK				0x3f
+#define     LTSSM_DETECT_QUIET			0x0
+#define     LTSSM_DETECT_ACTIVE			0x1
+#define     LTSSM_POLLING_ACTIVE		0x2
+#define     LTSSM_POLLING_COMPLIANCE		0x3
+#define     LTSSM_POLLING_CONFIGURATION		0x4
+#define     LTSSM_CONFIG_LINKWIDTH_START	0x5
+#define     LTSSM_CONFIG_LINKWIDTH_ACCEPT	0x6
+#define     LTSSM_CONFIG_LANENUM_ACCEPT		0x7
+#define     LTSSM_CONFIG_LANENUM_WAIT		0x8
+#define     LTSSM_CONFIG_COMPLETE		0x9
+#define     LTSSM_CONFIG_IDLE			0xa
+#define     LTSSM_RECOVERY_RCVR_LOCK		0xb
+#define     LTSSM_RECOVERY_SPEED		0xc
+#define     LTSSM_RECOVERY_RCVR_CFG		0xd
+#define     LTSSM_RECOVERY_IDLE			0xe
 #define     LTSSM_L0				0x10
+#define     LTSSM_RX_L0S_ENTRY			0x11
+#define     LTSSM_RX_L0S_IDLE			0x12
+#define     LTSSM_RX_L0S_FTS			0x13
+#define     LTSSM_TX_L0S_ENTRY			0x14
+#define     LTSSM_TX_L0S_IDLE			0x15
+#define     LTSSM_TX_L0S_FTS			0x16
+#define     LTSSM_L1_ENTRY			0x17
+#define     LTSSM_L1_IDLE			0x18
+#define     LTSSM_L2_IDLE			0x19
+#define     LTSSM_L2_TRANSMIT_WAKE		0x1a
+#define     LTSSM_DISABLED			0x20
+#define     LTSSM_LOOPBACK_ENTRY_MASTER		0x21
+#define     LTSSM_LOOPBACK_ACTIVE_MASTER	0x22
+#define     LTSSM_LOOPBACK_EXIT_MASTER		0x23
+#define     LTSSM_LOOPBACK_ENTRY_SLAVE		0x24
+#define     LTSSM_LOOPBACK_ACTIVE_SLAVE		0x25
+#define     LTSSM_LOOPBACK_EXIT_SLAVE		0x26
+#define     LTSSM_HOT_RESET			0x27
+#define     LTSSM_RECOVERY_EQUALIZATION_PHASE0	0x28
+#define     LTSSM_RECOVERY_EQUALIZATION_PHASE1	0x29
+#define     LTSSM_RECOVERY_EQUALIZATION_PHASE2	0x2a
+#define     LTSSM_RECOVERY_EQUALIZATION_PHASE3	0x2b
 #define     RC_BAR_CONFIG			0x300
 #define VENDOR_ID_REG				(LMI_BASE_ADDR + 0x44)
 
@@ -258,13 +295,35 @@ static inline u32 advk_readl(struct advk_pcie *pcie, u64 reg)
 	return readl(pcie->base + reg);
 }
 
-static int advk_pcie_link_up(struct advk_pcie *pcie)
+static u8 advk_pcie_ltssm_state(struct advk_pcie *pcie)
 {
-	u32 val, ltssm_state;
+	u32 val;
+	u8 ltssm_state;
 
 	val = advk_readl(pcie, CFG_REG);
 	ltssm_state = (val >> LTSSM_SHIFT) & LTSSM_MASK;
-	return ltssm_state >= LTSSM_L0;
+	return ltssm_state;
+}
+
+static inline bool advk_pcie_link_up(struct advk_pcie *pcie)
+{
+	/* check if LTSSM is in normal operation - some L* state */
+	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
+	return ltssm_state >= LTSSM_L0 && ltssm_state < LTSSM_DISABLED;
+}
+
+static inline bool advk_pcie_link_training(struct advk_pcie *pcie)
+{
+	/*
+	 * According to PCIe Base specification 3.0, Table 4-14: Link
+	 * Status Mapped to the LTSSM is Link Training mapped to LTSSM
+	 * Configuration and Recovery states.
+	 */
+	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
+	return ((ltssm_state >= LTSSM_CONFIG_LINKWIDTH_START &&
+		 ltssm_state < LTSSM_L0) ||
+		(ltssm_state >= LTSSM_RECOVERY_EQUALIZATION_PHASE0 &&
+		 ltssm_state <= LTSSM_RECOVERY_EQUALIZATION_PHASE3));
 }
 
 static int advk_pcie_wait_for_link(struct advk_pcie *pcie)
@@ -287,7 +346,7 @@ static void advk_pcie_wait_for_retrain(struct advk_pcie *pcie)
 	size_t retries;
 
 	for (retries = 0; retries < RETRAIN_WAIT_MAX_RETRIES; ++retries) {
-		if (!advk_pcie_link_up(pcie))
+		if (advk_pcie_link_training(pcie))
 			break;
 		udelay(RETRAIN_WAIT_USLEEP_US);
 	}
@@ -706,7 +765,7 @@ advk_pci_bridge_emul_pcie_conf_read(struct pci_bridge_emul *bridge,
 		/* u32 contains both PCI_EXP_LNKCTL and PCI_EXP_LNKSTA */
 		u32 val = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + reg) &
 			~(PCI_EXP_LNKSTA_LT << 16);
-		if (!advk_pcie_link_up(pcie))
+		if (advk_pcie_link_training(pcie))
 			val |= (PCI_EXP_LNKSTA_LT << 16);
 		*value = val;
 		return PCI_BRIDGE_EMUL_HANDLED;
-- 
2.32.0

[PATCH 13/13] PCI: aardvark: Fix reporting Data Link Layer Link Active

[Marek Behún]

From: Pali Rohár <pali@kernel.org>

Add support for reporting PCI_EXP_LNKSTA_DLLLA bit in Link Control register
on emulated bridge via current LTSSM state. Also correctly indicate DLLLA
capability via PCI_EXP_LNKCAP_DLLLARC bit in Link Control Capability
register.

Fixes: 8a3ebd8de328 ("PCI: aardvark: Implement emulated root PCI bridge config space")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Cc: stable@vger.kernel.org
---
 drivers/pci/controller/pci-aardvark.c | 29 ++++++++++++++++++++++++++-
 1 file changed, 28 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index 9465b630cede..f429c89ba4a1 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -312,6 +312,20 @@ static inline bool advk_pcie_link_up(struct advk_pcie *pcie)
 	return ltssm_state >= LTSSM_L0 && ltssm_state < LTSSM_DISABLED;
 }
 
+static inline bool advk_pcie_link_active(struct advk_pcie *pcie)
+{
+	/*
+	 * According to PCIe Base specification 3.0, Table 4-14: Link
+	 * Status Mapped to the LTSSM, and 4.2.6.3.6 Configuration.Idle
+	 * is Link Up mapped to LTSSM Configuration.Idle, Recovery, L0,
+	 * L0s, L1 and L2 states. And according to 3.2.1. Data Link
+	 * Control and Management State Machine Rules is DL Up status
+	 * reported in DL Active state.
+	 */
+	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
+	return ltssm_state >= LTSSM_CONFIG_IDLE && ltssm_state < LTSSM_DISABLED;
+}
+
 static inline bool advk_pcie_link_training(struct advk_pcie *pcie)
 {
 	/*
@@ -761,12 +775,26 @@ advk_pci_bridge_emul_pcie_conf_read(struct pci_bridge_emul *bridge,
 		return PCI_BRIDGE_EMUL_HANDLED;
 	}
 
+	case PCI_EXP_LNKCAP: {
+		u32 val = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + reg);
+		/*
+		 * PCI_EXP_LNKCAP_DLLLARC bit is hardwired in aardvark HW to 0.
+		 * But support for PCI_EXP_LNKSTA_DLLLA is emulated via ltssm
+		 * state so explicitly enable PCI_EXP_LNKCAP_DLLLARC flag.
+		 */
+		val |= PCI_EXP_LNKCAP_DLLLARC;
+		*value = val;
+		return PCI_BRIDGE_EMUL_HANDLED;
+	}
+
 	case PCI_EXP_LNKCTL: {
 		/* u32 contains both PCI_EXP_LNKCTL and PCI_EXP_LNKSTA */
 		u32 val = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + reg) &
 			~(PCI_EXP_LNKSTA_LT << 16);
 		if (advk_pcie_link_training(pcie))
 			val |= (PCI_EXP_LNKSTA_LT << 16);
+		if (advk_pcie_link_active(pcie))
+			val |= (PCI_EXP_LNKSTA_DLLLA << 16);
 		*value = val;
 		return PCI_BRIDGE_EMUL_HANDLED;
 	}
@@ -774,7 +802,6 @@ advk_pci_bridge_emul_pcie_conf_read(struct pci_bridge_emul *bridge,
 	case PCI_CAP_LIST_ID:
 	case PCI_EXP_DEVCAP:
 	case PCI_EXP_DEVCTL:
-	case PCI_EXP_LNKCAP:
 		*value = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + reg);
 		return PCI_BRIDGE_EMUL_HANDLED;
 	default:
-- 
2.32.0

Re: [PATCH 02/13] PCI: aardvark: Fix PCIe Max Payload Size setting

[Marek Behún]

Greg, Sasha,

I accidentally sent these patches to stable even though they are not
merged in upstream yet (by not configuring git-send-email correctly).

Sorry about that, please ignore this series in stable for now.

Marek

Re: [PATCH 12/13] PCI: aardvark: Fix checking for link up via LTSSM state

[Lorenzo Pieralisi]

On Fri, Oct 01, 2021 at 09:58:55PM +0200, Marek Behún wrote:
> From: Pali Rohár <pali@kernel.org>
> 
> Current implementation of advk_pcie_link_up() is wrong as it marks also
> link disabled or hot reset states as link up.
> 
> Fix it by marking link up only to those states which are defined in PCIe
> Base specification 3.0, Table 4-14: Link Status Mapped to the LTSSM.
> 
> To simplify implementation, Define macros for every LTSSM state which
> aardvark hardware can return in CFG_REG register.
> 
> Fix also checking for link training according to the same Table 4-14.
> Define a new function advk_pcie_link_training() for this purpose.
> 
> Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
> Signed-off-by: Pali Rohár <pali@kernel.org>
> Reviewed-by: Marek Behún <kabel@kernel.org>
> Signed-off-by: Marek Behún <kabel@kernel.org>
> Cc: Remi Pommarel <repk@triplefau.lt>
> Cc: stable@vger.kernel.org
> ---
>  drivers/pci/controller/pci-aardvark.c | 69 +++++++++++++++++++++++++--
>  1 file changed, 64 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
> index 5387d9cc3eba..9465b630cede 100644
> --- a/drivers/pci/controller/pci-aardvark.c
> +++ b/drivers/pci/controller/pci-aardvark.c
> @@ -165,7 +165,44 @@
>  #define CFG_REG					(LMI_BASE_ADDR + 0x0)
>  #define     LTSSM_SHIFT				24
>  #define     LTSSM_MASK				0x3f
> +#define     LTSSM_DETECT_QUIET			0x0
> +#define     LTSSM_DETECT_ACTIVE			0x1
> +#define     LTSSM_POLLING_ACTIVE		0x2
> +#define     LTSSM_POLLING_COMPLIANCE		0x3
> +#define     LTSSM_POLLING_CONFIGURATION		0x4
> +#define     LTSSM_CONFIG_LINKWIDTH_START	0x5
> +#define     LTSSM_CONFIG_LINKWIDTH_ACCEPT	0x6
> +#define     LTSSM_CONFIG_LANENUM_ACCEPT		0x7
> +#define     LTSSM_CONFIG_LANENUM_WAIT		0x8
> +#define     LTSSM_CONFIG_COMPLETE		0x9
> +#define     LTSSM_CONFIG_IDLE			0xa
> +#define     LTSSM_RECOVERY_RCVR_LOCK		0xb
> +#define     LTSSM_RECOVERY_SPEED		0xc
> +#define     LTSSM_RECOVERY_RCVR_CFG		0xd
> +#define     LTSSM_RECOVERY_IDLE			0xe
>  #define     LTSSM_L0				0x10
> +#define     LTSSM_RX_L0S_ENTRY			0x11
> +#define     LTSSM_RX_L0S_IDLE			0x12
> +#define     LTSSM_RX_L0S_FTS			0x13
> +#define     LTSSM_TX_L0S_ENTRY			0x14
> +#define     LTSSM_TX_L0S_IDLE			0x15
> +#define     LTSSM_TX_L0S_FTS			0x16
> +#define     LTSSM_L1_ENTRY			0x17
> +#define     LTSSM_L1_IDLE			0x18
> +#define     LTSSM_L2_IDLE			0x19
> +#define     LTSSM_L2_TRANSMIT_WAKE		0x1a
> +#define     LTSSM_DISABLED			0x20
> +#define     LTSSM_LOOPBACK_ENTRY_MASTER		0x21
> +#define     LTSSM_LOOPBACK_ACTIVE_MASTER	0x22
> +#define     LTSSM_LOOPBACK_EXIT_MASTER		0x23
> +#define     LTSSM_LOOPBACK_ENTRY_SLAVE		0x24
> +#define     LTSSM_LOOPBACK_ACTIVE_SLAVE		0x25
> +#define     LTSSM_LOOPBACK_EXIT_SLAVE		0x26
> +#define     LTSSM_HOT_RESET			0x27
> +#define     LTSSM_RECOVERY_EQUALIZATION_PHASE0	0x28
> +#define     LTSSM_RECOVERY_EQUALIZATION_PHASE1	0x29
> +#define     LTSSM_RECOVERY_EQUALIZATION_PHASE2	0x2a
> +#define     LTSSM_RECOVERY_EQUALIZATION_PHASE3	0x2b

An enum would be nicer IMO, for readability. Don't repost the series
yet.

Lorenzo

>  #define     RC_BAR_CONFIG			0x300
>  #define VENDOR_ID_REG				(LMI_BASE_ADDR + 0x44)
>  
> @@ -258,13 +295,35 @@ static inline u32 advk_readl(struct advk_pcie *pcie, u64 reg)
>  	return readl(pcie->base + reg);
>  }
>  
> -static int advk_pcie_link_up(struct advk_pcie *pcie)
> +static u8 advk_pcie_ltssm_state(struct advk_pcie *pcie)
>  {
> -	u32 val, ltssm_state;
> +	u32 val;
> +	u8 ltssm_state;
>  
>  	val = advk_readl(pcie, CFG_REG);
>  	ltssm_state = (val >> LTSSM_SHIFT) & LTSSM_MASK;
> -	return ltssm_state >= LTSSM_L0;
> +	return ltssm_state;
> +}
> +
> +static inline bool advk_pcie_link_up(struct advk_pcie *pcie)
> +{
> +	/* check if LTSSM is in normal operation - some L* state */
> +	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
> +	return ltssm_state >= LTSSM_L0 && ltssm_state < LTSSM_DISABLED;
> +}
> +
> +static inline bool advk_pcie_link_training(struct advk_pcie *pcie)
> +{
> +	/*
> +	 * According to PCIe Base specification 3.0, Table 4-14: Link
> +	 * Status Mapped to the LTSSM is Link Training mapped to LTSSM
> +	 * Configuration and Recovery states.
> +	 */
> +	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
> +	return ((ltssm_state >= LTSSM_CONFIG_LINKWIDTH_START &&
> +		 ltssm_state < LTSSM_L0) ||
> +		(ltssm_state >= LTSSM_RECOVERY_EQUALIZATION_PHASE0 &&
> +		 ltssm_state <= LTSSM_RECOVERY_EQUALIZATION_PHASE3));
>  }
>  
>  static int advk_pcie_wait_for_link(struct advk_pcie *pcie)
> @@ -287,7 +346,7 @@ static void advk_pcie_wait_for_retrain(struct advk_pcie *pcie)
>  	size_t retries;
>  
>  	for (retries = 0; retries < RETRAIN_WAIT_MAX_RETRIES; ++retries) {
> -		if (!advk_pcie_link_up(pcie))
> +		if (advk_pcie_link_training(pcie))
>  			break;
>  		udelay(RETRAIN_WAIT_USLEEP_US);
>  	}
> @@ -706,7 +765,7 @@ advk_pci_bridge_emul_pcie_conf_read(struct pci_bridge_emul *bridge,
>  		/* u32 contains both PCI_EXP_LNKCTL and PCI_EXP_LNKSTA */
>  		u32 val = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + reg) &
>  			~(PCI_EXP_LNKSTA_LT << 16);
> -		if (!advk_pcie_link_up(pcie))
> +		if (advk_pcie_link_training(pcie))
>  			val |= (PCI_EXP_LNKSTA_LT << 16);
>  		*value = val;
>  		return PCI_BRIDGE_EMUL_HANDLED;
> -- 
> 2.32.0
> 

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Lorenzo Pieralisi]

[+Marc - always better to have his eyes on IRQ handling code]

On Fri, Oct 01, 2021 at 09:58:49PM +0200, Marek Behún wrote:
> From: Pali Rohár <pali@kernel.org>
> 
> It is incorrect to clear status bits of masked interrupts.
> 
> The aardvark driver clears all status interrupt bits if no unmasked
> status bit is set. Masked bits should never be cleared.
> 
> Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
> Signed-off-by: Pali Rohár <pali@kernel.org>
> Reviewed-by: Marek Behún <kabel@kernel.org>
> Signed-off-by: Marek Behún <kabel@kernel.org>
> Cc: stable@vger.kernel.org
> ---
>  drivers/pci/controller/pci-aardvark.c | 5 +----
>  1 file changed, 1 insertion(+), 4 deletions(-)
> 
> diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
> index d5d6f92e5143..e4986806a189 100644
> --- a/drivers/pci/controller/pci-aardvark.c
> +++ b/drivers/pci/controller/pci-aardvark.c
> @@ -1295,11 +1295,8 @@ static void advk_pcie_handle_int(struct advk_pcie *pcie)
>  	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
>  	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
>  
> -	if (!isr0_status && !isr1_status) {
> -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);

This looks fine - on the other hand if no interrupt is set in the status
registers (that are filtered with the masks) we are dealing with a
spurious IRQ right ? Just gauging how severe this is.

Lorenzo

> +	if (!isr0_status && !isr1_status)
>  		return;
> -	}
>  
>  	/* Process MSI interrupts */
>  	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
> -- 
> 2.32.0
> 

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Marek Behún]

On Mon, 4 Oct 2021 15:06:53 +0100
Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote:

> [+Marc - always better to have his eyes on IRQ handling code]
> 
> > -	if (!isr0_status && !isr1_status) {
> > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);  
> 
> This looks fine - on the other hand if no interrupt is set in the status
> registers (that are filtered with the masks) we are dealing with a
> spurious IRQ right ? Just gauging how severe this is.

Yes, spurious IRQ can really happen.

Patch 7 in this series fixes an issue where aardvark does not mask all
interrupts, and then kernel can think that an interrupt is masked but
it really isn't.

Also, some interrupts may be masked by the user of the emulated bridge
(some other driver), so that they can be polled. But if we clear all of
them in the status, even the masked ones, then the other driver which
is polling will always get a zero.

Marek

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Marc Zyngier]

On Mon, 04 Oct 2021 15:06:53 +0100,
Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote:
> 
> [+Marc - always better to have his eyes on IRQ handling code]
> 
> On Fri, Oct 01, 2021 at 09:58:49PM +0200, Marek Behún wrote:
> > From: Pali Rohár <pali@kernel.org>
> > 
> > It is incorrect to clear status bits of masked interrupts.
> > 
> > The aardvark driver clears all status interrupt bits if no unmasked
> > status bit is set. Masked bits should never be cleared.
> > 
> > Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
> > Signed-off-by: Pali Rohár <pali@kernel.org>
> > Reviewed-by: Marek Behún <kabel@kernel.org>
> > Signed-off-by: Marek Behún <kabel@kernel.org>
> > Cc: stable@vger.kernel.org
> > ---
> >  drivers/pci/controller/pci-aardvark.c | 5 +----
> >  1 file changed, 1 insertion(+), 4 deletions(-)
> > 
> > diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
> > index d5d6f92e5143..e4986806a189 100644
> > --- a/drivers/pci/controller/pci-aardvark.c
> > +++ b/drivers/pci/controller/pci-aardvark.c
> > @@ -1295,11 +1295,8 @@ static void advk_pcie_handle_int(struct advk_pcie *pcie)
> >  	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
> >  	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
> >  
> > -	if (!isr0_status && !isr1_status) {
> > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
> 
> This looks fine - on the other hand if no interrupt is set in the status
> registers (that are filtered with the masks) we are dealing with a
> spurious IRQ right ? Just gauging how severe this is.
> 
> Lorenzo
> 
> > +	if (!isr0_status && !isr1_status)
> >  		return;

The whole thing is a bit odd. What the commit message doesn't say is
whether the status register shows the status of the line before
masking, or after masking.

The code seems to imply the former, but then the behaviour is
awkward. How did we end-up here the first place? if that's only a
spurious interrupt, then I'd probably simplify the code altogether,
and drop all the early return code. Something like below, as usual
completely untested.

	M.

diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
index 596ebcfcc82d..1d8f257ecb63 100644
--- a/drivers/pci/controller/pci-aardvark.c
+++ b/drivers/pci/controller/pci-aardvark.c
@@ -1275,7 +1275,8 @@ static void advk_pcie_handle_msi(struct advk_pcie *pcie)
 static void advk_pcie_handle_int(struct advk_pcie *pcie)
 {
 	u32 isr0_val, isr0_mask, isr0_status;
-	u32 isr1_val, isr1_mask, isr1_status;
+	u32 isr1_val, isr1_mask;
+	unsigned long isr1_status;
 	int i;
 
 	isr0_val = advk_readl(pcie, PCIE_ISR0_REG);
@@ -1285,22 +1286,14 @@ static void advk_pcie_handle_int(struct advk_pcie *pcie)
 	isr1_val = advk_readl(pcie, PCIE_ISR1_REG);
 	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
 	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
-
-	if (!isr0_status && !isr1_status) {
-		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
-		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
-		return;
-	}
+	isr1_status >> 8;
 
 	/* Process MSI interrupts */
 	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
 		advk_pcie_handle_msi(pcie);
 
 	/* Process legacy interrupts */
-	for (i = 0; i < PCI_NUM_INTX; i++) {
-		if (!(isr1_status & PCIE_ISR1_INTX_ASSERT(i)))
-			continue;
-
+	for_each_set_bit(i, &isr1_status, PCI_NUM_INTX) {
 		advk_writel(pcie, PCIE_ISR1_INTX_ASSERT(i),
 			    PCIE_ISR1_REG);
 

-- 
Without deviation from the norm, progress is not possible.

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Marek Behún]

On Mon, 04 Oct 2021 16:31:54 +0100
Marc Zyngier <maz@kernel.org> wrote:

> On Mon, 04 Oct 2021 15:06:53 +0100,
> Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote:
> > 
> > [+Marc - always better to have his eyes on IRQ handling code]
> > 
> > On Fri, Oct 01, 2021 at 09:58:49PM +0200, Marek Behún wrote:  
> > > From: Pali Rohár <pali@kernel.org>
> > > 
> > > It is incorrect to clear status bits of masked interrupts.
> > > 
> > > The aardvark driver clears all status interrupt bits if no
> > > unmasked status bit is set. Masked bits should never be cleared.
> > > 
> > > Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host
> > > controller driver") Signed-off-by: Pali Rohár <pali@kernel.org>
> > > Reviewed-by: Marek Behún <kabel@kernel.org>
> > > Signed-off-by: Marek Behún <kabel@kernel.org>
> > > Cc: stable@vger.kernel.org
> > > ---
> > >  drivers/pci/controller/pci-aardvark.c | 5 +----
> > >  1 file changed, 1 insertion(+), 4 deletions(-)
> > > 
> > > diff --git a/drivers/pci/controller/pci-aardvark.c
> > > b/drivers/pci/controller/pci-aardvark.c index
> > > d5d6f92e5143..e4986806a189 100644 ---
> > > a/drivers/pci/controller/pci-aardvark.c +++
> > > b/drivers/pci/controller/pci-aardvark.c @@ -1295,11 +1295,8 @@
> > > static void advk_pcie_handle_int(struct advk_pcie *pcie)
> > > isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG); isr1_status =
> > > isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK); 
> > > -	if (!isr0_status && !isr1_status) {
> > > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);  
> > 
> > This looks fine - on the other hand if no interrupt is set in the
> > status registers (that are filtered with the masks) we are dealing
> > with a spurious IRQ right ? Just gauging how severe this is.
> > 
> > Lorenzo
> >   
> > > +	if (!isr0_status && !isr1_status)
> > >  		return;  
> 
> The whole thing is a bit odd. What the commit message doesn't say is
> whether the status register shows the status of the line before
> masking, or after masking.

I don't quite understand what you are asking about.
If you are asking about the register itself:
the PCIE_ISR1_REG says which interrupts are currently set / active,
including those which are masked.

If you are asking about the isr1_status variable, it is the
status of the line after masking. I.e. masked interrupts are not set in
this variable, only active interrupts which are also unmasked. That is
obvious from the code.

> The code seems to imply the former, but then the behaviour is
> awkward. How did we end-up here the first place?

I answered this in reply to Lorenzo's comment on this patch, see
https://lore.kernel.org/linux-pci/20211004171823.0288684e@thinkpad/

> if that's only a
> spurious interrupt, then I'd probably simplify the code altogether,
> and drop all the early return code. Something like below, as usual
> completely untested.
> 
> 	M.
> 
> diff --git a/drivers/pci/controller/pci-aardvark.c
> b/drivers/pci/controller/pci-aardvark.c index
> 596ebcfcc82d..1d8f257ecb63 100644 ---
> a/drivers/pci/controller/pci-aardvark.c +++
> b/drivers/pci/controller/pci-aardvark.c @@ -1275,7 +1275,8 @@ static
> void advk_pcie_handle_msi(struct advk_pcie *pcie) static void
> advk_pcie_handle_int(struct advk_pcie *pcie) {
>  	u32 isr0_val, isr0_mask, isr0_status;
> -	u32 isr1_val, isr1_mask, isr1_status;
> +	u32 isr1_val, isr1_mask;
> +	unsigned long isr1_status;
>  	int i;
>  
>  	isr0_val = advk_readl(pcie, PCIE_ISR0_REG);
> @@ -1285,22 +1286,14 @@ static void advk_pcie_handle_int(struct
> advk_pcie *pcie) isr1_val = advk_readl(pcie, PCIE_ISR1_REG);
>  	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
>  	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
> -
> -	if (!isr0_status && !isr1_status) {
> -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
> -		return;
> -	}
> +	isr1_status >> 8;
>
>  	/* Process MSI interrupts */
>  	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
>  		advk_pcie_handle_msi(pcie);
>  
>  	/* Process legacy interrupts */
> -	for (i = 0; i < PCI_NUM_INTX; i++) {
> -		if (!(isr1_status & PCIE_ISR1_INTX_ASSERT(i)))
> -			continue;
> -
> +	for_each_set_bit(i, &isr1_status, PCI_NUM_INTX) {
>  		advk_writel(pcie, PCIE_ISR1_INTX_ASSERT(i),
>  			    PCIE_ISR1_REG);

1. what you are doing here is code cleanup. We are currently in the
   state where we have lots of fixes for this driver, which we are
   hoping will go also to stable. Some of them depend on these changes.
   Can we please first apply those fixes (we want to send them in
   batches to avoid sending 60 patchs in one series, since last time
   nobody wanted to review all of that) and do this afterwards?

2. you are throwing away lower 8 bits of isr1_status. We have follow-up
   patches (not in this series, but in another batch which we want to
   send after this) that will be using those lower 8 bits, so we do not
   want to throw away them now.

Marek

PS: You can look at all the prepared changes at
https://git.kernel.org/pub/scm/linux/kernel/git/pali/linux.git/log/?h=pci-aardvark

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Marc Zyngier]

On Tue, 05 Oct 2021 13:13:40 +0100,
Marek Behún <kabel@kernel.org> wrote:
> 
> On Mon, 04 Oct 2021 16:31:54 +0100
> Marc Zyngier <maz@kernel.org> wrote:
> 
> > On Mon, 04 Oct 2021 15:06:53 +0100,
> > Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote:
> > > 
> > > [+Marc - always better to have his eyes on IRQ handling code]
> > > 
> > > On Fri, Oct 01, 2021 at 09:58:49PM +0200, Marek Behún wrote:  
> > > > From: Pali Rohár <pali@kernel.org>
> > > > 
> > > > It is incorrect to clear status bits of masked interrupts.
> > > > 
> > > > The aardvark driver clears all status interrupt bits if no
> > > > unmasked status bit is set. Masked bits should never be cleared.
> > > > 
> > > > Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host
> > > > controller driver") Signed-off-by: Pali Rohár <pali@kernel.org>
> > > > Reviewed-by: Marek Behún <kabel@kernel.org>
> > > > Signed-off-by: Marek Behún <kabel@kernel.org>
> > > > Cc: stable@vger.kernel.org
> > > > ---
> > > >  drivers/pci/controller/pci-aardvark.c | 5 +----
> > > >  1 file changed, 1 insertion(+), 4 deletions(-)
> > > > 
> > > > diff --git a/drivers/pci/controller/pci-aardvark.c
> > > > b/drivers/pci/controller/pci-aardvark.c index
> > > > d5d6f92e5143..e4986806a189 100644 ---
> > > > a/drivers/pci/controller/pci-aardvark.c +++
> > > > b/drivers/pci/controller/pci-aardvark.c @@ -1295,11 +1295,8 @@
> > > > static void advk_pcie_handle_int(struct advk_pcie *pcie)
> > > > isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG); isr1_status =
> > > > isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK); 
> > > > -	if (!isr0_status && !isr1_status) {
> > > > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > > > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);  
> > > 
> > > This looks fine - on the other hand if no interrupt is set in the
> > > status registers (that are filtered with the masks) we are dealing
> > > with a spurious IRQ right ? Just gauging how severe this is.
> > > 
> > > Lorenzo
> > >   
> > > > +	if (!isr0_status && !isr1_status)
> > > >  		return;  
> > 
> > The whole thing is a bit odd. What the commit message doesn't say is
> > whether the status register shows the status of the line before
> > masking, or after masking.
> 
> I don't quite understand what you are asking about.
> If you are asking about the register itself:
> the PCIE_ISR1_REG says which interrupts are currently set / active,
> including those which are masked.

Then please say so in the commit message.

> If you are asking about the isr1_status variable, it is the
> status of the line after masking. I.e. masked interrupts are not set in
> this variable, only active interrupts which are also unmasked. That is
> obvious from the code.

Which is what I have said... two lines below. If you are going to
reply, please do so in context.

> 
> > The code seems to imply the former, but then the behaviour is
> > awkward. How did we end-up here the first place?
> 
> I answered this in reply to Lorenzo's comment on this patch, see
> https://lore.kernel.org/linux-pci/20211004171823.0288684e@thinkpad/

It did grace my inbox, thanks.

> > if that's only a
> > spurious interrupt, then I'd probably simplify the code altogether,
> > and drop all the early return code. Something like below, as usual
> > completely untested.
> > 
> > 	M.
> > 
> > diff --git a/drivers/pci/controller/pci-aardvark.c
> > b/drivers/pci/controller/pci-aardvark.c index
> > 596ebcfcc82d..1d8f257ecb63 100644 ---
> > a/drivers/pci/controller/pci-aardvark.c +++
> > b/drivers/pci/controller/pci-aardvark.c @@ -1275,7 +1275,8 @@ static
> > void advk_pcie_handle_msi(struct advk_pcie *pcie) static void
> > advk_pcie_handle_int(struct advk_pcie *pcie) {
> >  	u32 isr0_val, isr0_mask, isr0_status;
> > -	u32 isr1_val, isr1_mask, isr1_status;
> > +	u32 isr1_val, isr1_mask;
> > +	unsigned long isr1_status;
> >  	int i;
> >  
> >  	isr0_val = advk_readl(pcie, PCIE_ISR0_REG);
> > @@ -1285,22 +1286,14 @@ static void advk_pcie_handle_int(struct
> > advk_pcie *pcie) isr1_val = advk_readl(pcie, PCIE_ISR1_REG);
> >  	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
> >  	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
> > -
> > -	if (!isr0_status && !isr1_status) {
> > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
> > -		return;
> > -	}
> > +	isr1_status >> 8;
> >
> >  	/* Process MSI interrupts */
> >  	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
> >  		advk_pcie_handle_msi(pcie);
> >  
> >  	/* Process legacy interrupts */
> > -	for (i = 0; i < PCI_NUM_INTX; i++) {
> > -		if (!(isr1_status & PCIE_ISR1_INTX_ASSERT(i)))
> > -			continue;
> > -
> > +	for_each_set_bit(i, &isr1_status, PCI_NUM_INTX) {
> >  		advk_writel(pcie, PCIE_ISR1_INTX_ASSERT(i),
> >  			    PCIE_ISR1_REG);
> 
> 1. what you are doing here is code cleanup. We are currently in the
>    state where we have lots of fixes for this driver, which we are
>    hoping will go also to stable.

Yes, it is code cleanup. Because I don't find this patch to be very
good, TBH. As for going into stable, that's not relevant for this
discussion.

>    Some of them depend on these changes.
>    Can we please first apply those fixes (we want to send them in
>    batches to avoid sending 60 patchs in one series, since last time
>    nobody wanted to review all of that) and do this afterwards?

It would be better to start with patches that are in a better
shape. After all, this is what the code review process is about. This
isn't "just take my patches".

> 2. you are throwing away lower 8 bits of isr1_status. We have follow-up
>    patches (not in this series, but in another batch which we want to
>    send after this) that will be using those lower 8 bits, so we do not
>    want to throw away them now.

I'm discarding these bits because *in isolation*, that's the correct
thing to do. Feel free to propose a better patch that doesn't discard
these bits and still makes the code more palatable.

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Pali Rohár]

On Tuesday 05 October 2021 13:42:02 Marc Zyngier wrote:
> On Tue, 05 Oct 2021 13:13:40 +0100,
> Marek Behún <kabel@kernel.org> wrote:
> > 
> > On Mon, 04 Oct 2021 16:31:54 +0100
> > Marc Zyngier <maz@kernel.org> wrote:
> > 
> > > On Mon, 04 Oct 2021 15:06:53 +0100,
> > > Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote:
> > > > 
> > > > [+Marc - always better to have his eyes on IRQ handling code]
> > > > 
> > > > On Fri, Oct 01, 2021 at 09:58:49PM +0200, Marek Behún wrote:  
> > > > > From: Pali Rohár <pali@kernel.org>
> > > > > 
> > > > > It is incorrect to clear status bits of masked interrupts.
> > > > > 
> > > > > The aardvark driver clears all status interrupt bits if no
> > > > > unmasked status bit is set. Masked bits should never be cleared.
> > > > > 
> > > > > Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host
> > > > > controller driver") Signed-off-by: Pali Rohár <pali@kernel.org>
> > > > > Reviewed-by: Marek Behún <kabel@kernel.org>
> > > > > Signed-off-by: Marek Behún <kabel@kernel.org>
> > > > > Cc: stable@vger.kernel.org
> > > > > ---
> > > > >  drivers/pci/controller/pci-aardvark.c | 5 +----
> > > > >  1 file changed, 1 insertion(+), 4 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/pci/controller/pci-aardvark.c
> > > > > b/drivers/pci/controller/pci-aardvark.c index
> > > > > d5d6f92e5143..e4986806a189 100644 ---
> > > > > a/drivers/pci/controller/pci-aardvark.c +++
> > > > > b/drivers/pci/controller/pci-aardvark.c @@ -1295,11 +1295,8 @@
> > > > > static void advk_pcie_handle_int(struct advk_pcie *pcie)
> > > > > isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG); isr1_status =
> > > > > isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK); 
> > > > > -	if (!isr0_status && !isr1_status) {
> > > > > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > > > > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);  
> > > > 
> > > > This looks fine - on the other hand if no interrupt is set in the
> > > > status registers (that are filtered with the masks) we are dealing
> > > > with a spurious IRQ right ? Just gauging how severe this is.
> > > > 
> > > > Lorenzo
> > > >   
> > > > > +	if (!isr0_status && !isr1_status)
> > > > >  		return;  
> > > 
> > > The whole thing is a bit odd. What the commit message doesn't say is
> > > whether the status register shows the status of the line before
> > > masking, or after masking.
> > 
> > I don't quite understand what you are asking about.
> > If you are asking about the register itself:
> > the PCIE_ISR1_REG says which interrupts are currently set / active,
> > including those which are masked.
> 
> Then please say so in the commit message.

Very well, we shall do so.

> > If you are asking about the isr1_status variable, it is the
> > status of the line after masking. I.e. masked interrupts are not set in
> > this variable, only active interrupts which are also unmasked. That is
> > obvious from the code.
> 
> Which is what I have said... two lines below. If you are going to
> reply, please do so in context.
> 
> > 
> > > The code seems to imply the former, but then the behaviour is
> > > awkward. How did we end-up here the first place?
> > 
> > I answered this in reply to Lorenzo's comment on this patch, see
> > https://lore.kernel.org/linux-pci/20211004171823.0288684e@thinkpad/
> 
> It did grace my inbox, thanks.
> 
> > > if that's only a
> > > spurious interrupt, then I'd probably simplify the code altogether,
> > > and drop all the early return code. Something like below, as usual
> > > completely untested.
> > > 
> > > 	M.
> > > 
> > > diff --git a/drivers/pci/controller/pci-aardvark.c
> > > b/drivers/pci/controller/pci-aardvark.c index
> > > 596ebcfcc82d..1d8f257ecb63 100644 ---
> > > a/drivers/pci/controller/pci-aardvark.c +++
> > > b/drivers/pci/controller/pci-aardvark.c @@ -1275,7 +1275,8 @@ static
> > > void advk_pcie_handle_msi(struct advk_pcie *pcie) static void
> > > advk_pcie_handle_int(struct advk_pcie *pcie) {
> > >  	u32 isr0_val, isr0_mask, isr0_status;
> > > -	u32 isr1_val, isr1_mask, isr1_status;
> > > +	u32 isr1_val, isr1_mask;
> > > +	unsigned long isr1_status;
> > >  	int i;
> > >  
> > >  	isr0_val = advk_readl(pcie, PCIE_ISR0_REG);
> > > @@ -1285,22 +1286,14 @@ static void advk_pcie_handle_int(struct
> > > advk_pcie *pcie) isr1_val = advk_readl(pcie, PCIE_ISR1_REG);
> > >  	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
> > >  	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
> > > -
> > > -	if (!isr0_status && !isr1_status) {
> > > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
> > > -		return;
> > > -	}
> > > +	isr1_status >> 8;

Hello!

I dislike this approach. It adds another magic number which is just
causing issues. Please read commit message for patch 11/13 where we
describe why such magic constants are bad and already caused lot of
issues in this driver.

> > >  	/* Process MSI interrupts */
> > >  	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
> > >  		advk_pcie_handle_msi(pcie);
> > >  
> > >  	/* Process legacy interrupts */
> > > -	for (i = 0; i < PCI_NUM_INTX; i++) {
> > > -		if (!(isr1_status & PCIE_ISR1_INTX_ASSERT(i)))
> > > -			continue;
> > > -
> > > +	for_each_set_bit(i, &isr1_status, PCI_NUM_INTX) {
> > >  		advk_writel(pcie, PCIE_ISR1_INTX_ASSERT(i),
> > >  			    PCIE_ISR1_REG);
> > 
> > 1. what you are doing here is code cleanup. We are currently in the
> >    state where we have lots of fixes for this driver, which we are
> >    hoping will go also to stable.
> 
> Yes, it is code cleanup. Because I don't find this patch to be very
> good, TBH. As for going into stable, that's not relevant for this
> discussion.
> 
> >    Some of them depend on these changes.
> >    Can we please first apply those fixes (we want to send them in
> >    batches to avoid sending 60 patchs in one series, since last time
> >    nobody wanted to review all of that) and do this afterwards?
> 
> It would be better to start with patches that are in a better
> shape. After all, this is what the code review process is about. This
> isn't "just take my patches".
> 
> > 2. you are throwing away lower 8 bits of isr1_status. We have follow-up
> >    patches (not in this series, but in another batch which we want to
> >    send after this) that will be using those lower 8 bits, so we do not
> >    want to throw away them now.
> 
> I'm discarding these bits because *in isolation*, that's the correct
> thing to do. Feel free to propose a better patch that doesn't discard
> these bits and still makes the code more palatable.

The code pattern in this function is: compose irs*_status variable and
then compare it with register macros defined at the top of driver. Each
bit in this register represent some event and for each event there is
simple macro to match.

So with your proposed change it would break all macros (as they are
going to be shifted by magic constant) and then this code disallow
access to events represented by low bits. And also it makes code pattern
different for isr0_status and isr1_status variables which is very
confusing and probably source for introduction of new bugs.

Also the whole early-return optimization can be removed as it does not
change functionality. So we will do so.

But we do not agree with the lower 8 bit discard of the isr1_status
variable as explained above.

So if we add the explanation to commit message and drop the early
return, would it be ok?

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Marc Zyngier]

On Tue, 05 Oct 2021 14:15:45 +0100,
Pali Rohár <pali@kernel.org> wrote:
> 
> Hello!
> 
> I dislike this approach. It adds another magic number which is just
> causing issues. Please read commit message for patch 11/13 where we
> describe why such magic constants are bad and already caused lot of
> issues in this driver.

As I said, feel free to write something better.

> 
> > > >  	/* Process MSI interrupts */
> > > >  	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
> > > >  		advk_pcie_handle_msi(pcie);
> > > >  
> > > >  	/* Process legacy interrupts */
> > > > -	for (i = 0; i < PCI_NUM_INTX; i++) {
> > > > -		if (!(isr1_status & PCIE_ISR1_INTX_ASSERT(i)))
> > > > -			continue;
> > > > -
> > > > +	for_each_set_bit(i, &isr1_status, PCI_NUM_INTX) {
> > > >  		advk_writel(pcie, PCIE_ISR1_INTX_ASSERT(i),
> > > >  			    PCIE_ISR1_REG);
> > > 
> > > 1. what you are doing here is code cleanup. We are currently in the
> > >    state where we have lots of fixes for this driver, which we are
> > >    hoping will go also to stable.
> > 
> > Yes, it is code cleanup. Because I don't find this patch to be very
> > good, TBH. As for going into stable, that's not relevant for this
> > discussion.
> > 
> > >    Some of them depend on these changes.
> > >    Can we please first apply those fixes (we want to send them in
> > >    batches to avoid sending 60 patchs in one series, since last time
> > >    nobody wanted to review all of that) and do this afterwards?
> > 
> > It would be better to start with patches that are in a better
> > shape. After all, this is what the code review process is about. This
> > isn't "just take my patches".
> > 
> > > 2. you are throwing away lower 8 bits of isr1_status. We have follow-up
> > >    patches (not in this series, but in another batch which we want to
> > >    send after this) that will be using those lower 8 bits, so we do not
> > >    want to throw away them now.
> > 
> > I'm discarding these bits because *in isolation*, that's the correct
> > thing to do. Feel free to propose a better patch that doesn't discard
> > these bits and still makes the code more palatable.
> 
> The code pattern in this function is: compose irs*_status variable and
> then compare it with register macros defined at the top of driver. Each
> bit in this register represent some event and for each event there is
> simple macro to match.
> 
> So with your proposed change it would break all macros (as they are
> going to be shifted by magic constant) and then this code disallow
> access to events represented by low bits. And also it makes code pattern
> different for isr0_status and isr1_status variables which is very
> confusing and probably source for introduction of new bugs.

Read what I have said: I'm suggesting changes based on this patch *in
isolation*. I don't see any other related patch in my inbox (nor do I
want to receive any). Feel free to suggest something better (that's
the third time I write this...).

> Also the whole early-return optimization can be removed as it does not
> change functionality. So we will do so.
> 
> But we do not agree with the lower 8 bit discard of the isr1_status
> variable as explained above.
> 
> So if we add the explanation to commit message and drop the early
> return, would it be ok?

Do what you want. I have no interest in this particular piece of code,
and only replied to Lorenzo's request. It doesn't change what I think
of this patch, but I have too much on my plate to get dragged into
sterile arguments.

	M.

-- 
Without deviation from the norm, progress is not possible.

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Lorenzo Pieralisi]

On Tue, Oct 05, 2021 at 03:15:45PM +0200, Pali Rohár wrote:
> On Tuesday 05 October 2021 13:42:02 Marc Zyngier wrote:
> > On Tue, 05 Oct 2021 13:13:40 +0100,
> > Marek Behún <kabel@kernel.org> wrote:
> > > 
> > > On Mon, 04 Oct 2021 16:31:54 +0100
> > > Marc Zyngier <maz@kernel.org> wrote:
> > > 
> > > > On Mon, 04 Oct 2021 15:06:53 +0100,
> > > > Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote:
> > > > > 
> > > > > [+Marc - always better to have his eyes on IRQ handling code]
> > > > > 
> > > > > On Fri, Oct 01, 2021 at 09:58:49PM +0200, Marek Behún wrote:  
> > > > > > From: Pali Rohár <pali@kernel.org>
> > > > > > 
> > > > > > It is incorrect to clear status bits of masked interrupts.
> > > > > > 
> > > > > > The aardvark driver clears all status interrupt bits if no
> > > > > > unmasked status bit is set. Masked bits should never be cleared.
> > > > > > 
> > > > > > Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host
> > > > > > controller driver") Signed-off-by: Pali Rohár <pali@kernel.org>
> > > > > > Reviewed-by: Marek Behún <kabel@kernel.org>
> > > > > > Signed-off-by: Marek Behún <kabel@kernel.org>
> > > > > > Cc: stable@vger.kernel.org
> > > > > > ---
> > > > > >  drivers/pci/controller/pci-aardvark.c | 5 +----
> > > > > >  1 file changed, 1 insertion(+), 4 deletions(-)
> > > > > > 
> > > > > > diff --git a/drivers/pci/controller/pci-aardvark.c
> > > > > > b/drivers/pci/controller/pci-aardvark.c index
> > > > > > d5d6f92e5143..e4986806a189 100644 ---
> > > > > > a/drivers/pci/controller/pci-aardvark.c +++
> > > > > > b/drivers/pci/controller/pci-aardvark.c @@ -1295,11 +1295,8 @@
> > > > > > static void advk_pcie_handle_int(struct advk_pcie *pcie)
> > > > > > isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG); isr1_status =
> > > > > > isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK); 
> > > > > > -	if (!isr0_status && !isr1_status) {
> > > > > > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > > > > > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);  
> > > > > 
> > > > > This looks fine - on the other hand if no interrupt is set in the
> > > > > status registers (that are filtered with the masks) we are dealing
> > > > > with a spurious IRQ right ? Just gauging how severe this is.
> > > > > 
> > > > > Lorenzo
> > > > >   
> > > > > > +	if (!isr0_status && !isr1_status)
> > > > > >  		return;  
> > > > 
> > > > The whole thing is a bit odd. What the commit message doesn't say is
> > > > whether the status register shows the status of the line before
> > > > masking, or after masking.
> > > 
> > > I don't quite understand what you are asking about.
> > > If you are asking about the register itself:
> > > the PCIE_ISR1_REG says which interrupts are currently set / active,
> > > including those which are masked.
> > 
> > Then please say so in the commit message.
> 
> Very well, we shall do so.
> 
> > > If you are asking about the isr1_status variable, it is the
> > > status of the line after masking. I.e. masked interrupts are not set in
> > > this variable, only active interrupts which are also unmasked. That is
> > > obvious from the code.
> > 
> > Which is what I have said... two lines below. If you are going to
> > reply, please do so in context.
> > 
> > > 
> > > > The code seems to imply the former, but then the behaviour is
> > > > awkward. How did we end-up here the first place?
> > > 
> > > I answered this in reply to Lorenzo's comment on this patch, see
> > > https://lore.kernel.org/linux-pci/20211004171823.0288684e@thinkpad/
> > 
> > It did grace my inbox, thanks.
> > 
> > > > if that's only a
> > > > spurious interrupt, then I'd probably simplify the code altogether,
> > > > and drop all the early return code. Something like below, as usual
> > > > completely untested.
> > > > 
> > > > 	M.
> > > > 
> > > > diff --git a/drivers/pci/controller/pci-aardvark.c
> > > > b/drivers/pci/controller/pci-aardvark.c index
> > > > 596ebcfcc82d..1d8f257ecb63 100644 ---
> > > > a/drivers/pci/controller/pci-aardvark.c +++
> > > > b/drivers/pci/controller/pci-aardvark.c @@ -1275,7 +1275,8 @@ static
> > > > void advk_pcie_handle_msi(struct advk_pcie *pcie) static void
> > > > advk_pcie_handle_int(struct advk_pcie *pcie) {
> > > >  	u32 isr0_val, isr0_mask, isr0_status;
> > > > -	u32 isr1_val, isr1_mask, isr1_status;
> > > > +	u32 isr1_val, isr1_mask;
> > > > +	unsigned long isr1_status;
> > > >  	int i;
> > > >  
> > > >  	isr0_val = advk_readl(pcie, PCIE_ISR0_REG);
> > > > @@ -1285,22 +1286,14 @@ static void advk_pcie_handle_int(struct
> > > > advk_pcie *pcie) isr1_val = advk_readl(pcie, PCIE_ISR1_REG);
> > > >  	isr1_mask = advk_readl(pcie, PCIE_ISR1_MASK_REG);
> > > >  	isr1_status = isr1_val & ((~isr1_mask) & PCIE_ISR1_ALL_MASK);
> > > > -
> > > > -	if (!isr0_status && !isr1_status) {
> > > > -		advk_writel(pcie, isr0_val, PCIE_ISR0_REG);
> > > > -		advk_writel(pcie, isr1_val, PCIE_ISR1_REG);
> > > > -		return;
> > > > -	}
> > > > +	isr1_status >> 8;
> 
> Hello!
> 
> I dislike this approach. It adds another magic number which is just
> causing issues. Please read commit message for patch 11/13 where we
> describe why such magic constants are bad and already caused lot of
> issues in this driver.
> 
> > > >  	/* Process MSI interrupts */
> > > >  	if (isr0_status & PCIE_ISR0_MSI_INT_PENDING)
> > > >  		advk_pcie_handle_msi(pcie);
> > > >  
> > > >  	/* Process legacy interrupts */
> > > > -	for (i = 0; i < PCI_NUM_INTX; i++) {
> > > > -		if (!(isr1_status & PCIE_ISR1_INTX_ASSERT(i)))
> > > > -			continue;
> > > > -
> > > > +	for_each_set_bit(i, &isr1_status, PCI_NUM_INTX) {
> > > >  		advk_writel(pcie, PCIE_ISR1_INTX_ASSERT(i),
> > > >  			    PCIE_ISR1_REG);
> > > 
> > > 1. what you are doing here is code cleanup. We are currently in the
> > >    state where we have lots of fixes for this driver, which we are
> > >    hoping will go also to stable.
> > 
> > Yes, it is code cleanup. Because I don't find this patch to be very
> > good, TBH. As for going into stable, that's not relevant for this
> > discussion.
> > 
> > >    Some of them depend on these changes.
> > >    Can we please first apply those fixes (we want to send them in
> > >    batches to avoid sending 60 patchs in one series, since last time
> > >    nobody wanted to review all of that) and do this afterwards?
> > 
> > It would be better to start with patches that are in a better
> > shape. After all, this is what the code review process is about. This
> > isn't "just take my patches".
> > 
> > > 2. you are throwing away lower 8 bits of isr1_status. We have follow-up
> > >    patches (not in this series, but in another batch which we want to
> > >    send after this) that will be using those lower 8 bits, so we do not
> > >    want to throw away them now.
> > 
> > I'm discarding these bits because *in isolation*, that's the correct
> > thing to do. Feel free to propose a better patch that doesn't discard
> > these bits and still makes the code more palatable.
> 
> The code pattern in this function is: compose irs*_status variable and
> then compare it with register macros defined at the top of driver. Each
> bit in this register represent some event and for each event there is
> simple macro to match.
> 
> So with your proposed change it would break all macros (as they are
> going to be shifted by magic constant) and then this code disallow
> access to events represented by low bits. And also it makes code pattern
> different for isr0_status and isr1_status variables which is very
> confusing and probably source for introduction of new bugs.
> 
> Also the whole early-return optimization can be removed as it does not
> change functionality. So we will do so.
> 
> But we do not agree with the lower 8 bit discard of the isr1_status
> variable as explained above.
> 
> So if we add the explanation to commit message and drop the early
> return, would it be ok?

Please repost a v2 of this series with the review comments you
received (and tags removed if they are not applicable as things
stand in mainline) and we will take it from there.

Thanks,
Lorenzo

Re: [PATCH 06/13] PCI: aardvark: Do not clear status bits of masked interrupts

[Pali Rohár]

On Tuesday 05 October 2021 16:42:29 Marc Zyngier wrote:
> As I said, feel free to write something better.

Marek now sent a new version, I hope it is better:
https://lore.kernel.org/linux-pci/20211005180952.6812-7-kabel@kernel.org/


Anyway, I was thinking more if it is possible to end up in state when
*_status variables are zero also after applying patch 7/13.

I do not know how precisely are bits 8-11 of ISR1 reg handled in HW as
description is completely missing in all documentations which I read.
These bits represents events for legacy INTx interrupts.

And maybe it is possible that driver for endpoint card run some function
in context when all CPU interrupts are disabled and do something which
cause card to send Assert_INTA message followed by Deassert_INTA (e.g.
poke some register and immediately process event which deassert intx).
After endpoint driver function finish its execution then CPU receive
PCIe summary interrupt and pci controller driver would see that no event
is pending and noting to process.