Skip to contentSkip to conversation

drm/bridge: tc358767: Increase AUX transfer length limit

PATCH New Sasha Levin 1 participant 28 messages Sep 24, 2019

Subsystems (8)
lore
Share & tools
lore
author maintainer
#3 lore raw

PATCH 4/28 video: ssd1307fb: Start page range at page_offset

In reply to Sasha Levin

From: Marko Kohtala <[email protected]>

[ Upstream commit dd9782834dd9dde3624ff1acea8859f3d3e792d4 ]

The page_offset was only applied to the end of the page range. This caused the display updates to cause a scrolling effect on the display because the amount of data written to the display did not match the range display expected.

Fixes 301bc0675b67 ("video: ssd1307fb: Make use of horizontal addressing mode")
Cc Mark Rutland <[email protected]>
Cc Rob Herring <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc David Airlie <[email protected]>
Cc Michal Vokáč <[email protected]>
---
 drivers/video/fbdev/ssd1307fb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/video/fbdev/ssd1307fb.c +1 -1
index f599520374ddf..5f7dbf1c46092 100644
@@ -433,7 +433,7 @@ static int ssd1307fb_init(struct ssd1307fb_par *par)
 	if (ret < 0)
 		return ret;
 
-	ret = ssd1307fb_write_cmd(par->client, 0x0);
+	ret = ssd1307fb_write_cmd(par->client, par->page_offset);
 	if (ret < 0)
 		return ret;
 
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 5/28 drm/stm: attach gem fence to atomic state

In reply to Sasha Levin

From: Ahmad Fatoum <[email protected]>

[ Upstream commit 8fabc9c3109a71b3577959a05408153ae69ccd8d ]

To properly synchronize with other devices the fence from the GEM object backing the framebuffer needs to be attached to the atomic state, so the commit work can wait on fence signaling.

---
 drivers/gpu/drm/stm/ltdc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/gpu/drm/stm/ltdc.c +2 -0
index d394a03632c45..c3bd80b03f165 100644
@@ -20,6 +20,7 @@
 #include <drm/drm_crtc_helper.h>
 #include <drm/drm_fb_cma_helper.h>
 #include <drm/drm_gem_cma_helper.h>
+#include <drm/drm_gem_framebuffer_helper.h>
 #include <drm/drm_of.h>
 #include <drm/drm_bridge.h>
 #include <drm/drm_plane_helper.h>
@@ -691,6 +692,7 @@ static const struct drm_plane_funcs ltdc_plane_funcs = {
 };
 
 static const struct drm_plane_helper_funcs ltdc_plane_helper_funcs = {
+	.prepare_fb = drm_gem_fb_prepare_fb,
 	.atomic_check = ltdc_plane_atomic_check,
 	.atomic_update = ltdc_plane_atomic_update,
 	.atomic_disable = ltdc_plane_atomic_disable,
Signature · 1 line
2.20.1
author maintainer
#5 lore raw

PATCH 6/28 drm/radeon: Fix EEH during kexec

In reply to Sasha Levin

From: KyleMahlkuch <[email protected]>

[ Upstream commit 6f7fe9a93e6c09bf988c5059403f5f88e17e21e6 ]

During kexec some adapters hit an EEH since they are not properly shut down in the radeon_pci_shutdown() function. Adding radeon_suspend_kms() fixes this issue.

---
 drivers/gpu/drm/radeon/radeon_drv.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpu/drm/radeon/radeon_drv.c +8 -0
index f4becad0a78c0..54d97dd5780a1 100644
@@ -368,11 +368,19 @@ radeon_pci_remove(struct pci_dev *pdev)
 static void
 radeon_pci_shutdown(struct pci_dev *pdev)
 {
+	struct drm_device *ddev = pci_get_drvdata(pdev);
+
 	/* if we are running in a VM, make sure the device
 	 * torn down properly on reboot/shutdown
 	 */
 	if (radeon_device_is_virtual())
 		radeon_pci_remove(pdev);
+
+	/* Some adapters need to be suspended before a
+	* shutdown occurs in order to prevent an error
+	* during kexec.
+	*/
+	radeon_suspend_kms(ddev, true, true, false);
 }
 
 static int radeon_pmops_suspend(struct device *dev)
Signature · 6 lines
2.20.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author maintainer
#6 lore raw

PATCH 7/28 gpu: drm: radeon: Fix a possible null-pointer dereference in radeon_connector_set_property()

In reply to Sasha Levin

From: Jia-Ju Bai <[email protected]>

[ Upstream commit f3eb9b8f67bc28783eddc142ad805ebdc53d6339 ]

In radeon_connector_set_property(), there is an if statement on line 743 to check whether connector->encoder is NULL:

    if (connector->encoder)

When connector->encoder is NULL, it is used on line 755:

    if (connector->encoder->crtc)

Thus, a possible null-pointer dereference may occur.

To fix this bug, connector->encoder is checked before being used.

This bug is found by a static analysis tool STCheck written by us.

---
 drivers/gpu/drm/radeon/radeon_connectors.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/radeon/radeon_connectors.c +1 -1
index 337d3a1c2a409..48f752cf7a920 100644
@@ -764,7 +764,7 @@ static int radeon_connector_set_property(struct drm_connector *connector, struct
 
 		radeon_encoder->output_csc = val;
 
-		if (connector->encoder->crtc) {
+		if (connector->encoder && connector->encoder->crtc) {
 			struct drm_crtc *crtc  = connector->encoder->crtc;
 			struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
 
Signature · 6 lines
2.20.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author maintainer
#10 lore raw

PATCH 11/28 drm/amd/powerplay/smu7: enforce minimal VBITimeout (v2)

In reply to Sasha Levin

From: Ahzo <[email protected]>

[ Upstream commit f659bb6dae58c113805f92822e4c16ddd3156b79 ]

This fixes screen corruption/flickering on 75 Hz displays.

v2: make print statement debug only (Alex)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102646

---
 drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c +5 -0
index 336fdd8c7db08..61141bc3edfe9 100644
@@ -3972,6 +3972,11 @@ static int smu7_program_display_gap(struct pp_hwmgr *hwmgr)
 
 	data->frame_time_x2 = frame_time_in_us * 2 / 100;
 
+	if (data->frame_time_x2 < 280) {
+		pr_debug("%s: enforce minimal VBITimeout: %d -> 280\n", __func__, data->frame_time_x2);
+		data->frame_time_x2 = 280;
+	}
+
 	display_gap2 = pre_vbi_time_in_us * (ref_clock / 100);
 
 	cgs_write_ind_register(hwmgr->device, CGS_IND_REG__SMC, ixCG_DISPLAY_GAP_CNTL2, display_gap2);
Signature · 6 lines
2.20.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author maintainer
#18 lore raw

PATCH 19/28 pinctrl: tegra: Fix write barrier placement in pmx_writel

In reply to Sasha Levin

From: Sowjanya Komatineni <[email protected]>

[ Upstream commit c2cf351eba2ff6002ce8eb178452219d2521e38e ]

pmx_writel uses writel which inserts write barrier before the register write.

This patch has fix to replace writel with writel_relaxed followed by a readback and memory barrier to ensure write operation is completed for successful pinctrl change.

---
 drivers/pinctrl/tegra/pinctrl-tegra.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/pinctrl/tegra/pinctrl-tegra.c +3 -1
index 51716819129d2..e5c9b9c684289 100644
@@ -51,7 +51,9 @@ static inline u32 pmx_readl(struct tegra_pmx *pmx, u32 bank, u32 reg)
 
 static inline void pmx_writel(struct tegra_pmx *pmx, u32 val, u32 bank, u32 reg)
 {
-	writel(val, pmx->regs[bank] + reg);
+	writel_relaxed(val, pmx->regs[bank] + reg);
+	/* make sure pinmux register write completed */
+	pmx_readl(pmx, bank, reg);
 }
 
 static int tegra_pinctrl_get_groups_count(struct pinctrl_dev *pctldev)
Signature · 1 line
2.20.1
author maintainer
#20 lore raw

PATCH 21/28 drm/nouveau/volt: Fix for some cards having 0 maximum voltage

In reply to Sasha Levin

From: Mark Menzynski <[email protected]>

[ Upstream commit a1af2afbd244089560794c260b2d4326a86e39b6 ]

Some, mostly Fermi, vbioses appear to have zero max voltage. That causes Nouveau to not parse voltage entries, thus users not being able to set higher clocks.

When changing this value Nvidia driver still appeared to ignore it, and I wasn't able to find out why, thus the code is ignoring the value if it is zero.

---
 drivers/gpu/drm/nouveau/nvkm/subdev/bios/volt.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/gpu/drm/nouveau/nvkm/subdev/bios/volt.c +2 -0
index 7143ea4611aa3..33a9fb5ac5585 100644
@@ -96,6 +96,8 @@ nvbios_volt_parse(struct nvkm_bios *bios, u8 *ver, u8 *hdr, u8 *cnt, u8 *len,
 		info->min     = min(info->base,
 				    info->base + info->step * info->vidmask);
 		info->max     = nvbios_rd32(bios, volt + 0x0e);
+		if (!info->max)
+			info->max = max(info->base, info->base + info->step * info->vidmask);
 		break;
 	case 0x50:
 		info->min     = nvbios_rd32(bios, volt + 0x0a);
Signature · 1 line
2.20.1
author maintainer
#21 lore raw

PATCH 22/28 drm/amdgpu/si: fix ASIC tests

In reply to Sasha Levin

From: Jean Delvare <[email protected]>

[ Upstream commit 77efe48a729588527afb4d5811b9e0acb29f5e51 ]

Comparing adev->family with CHIP constants is not correct. adev->family can only be compared with AMDGPU_FAMILY constants and adev->asic_type is the struct member to compare with CHIP constants. They are separate identification spaces.

Fixes 62a37553414a ("drm/amdgpu: add si implementation v10")
Cc Ken Wang <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc "David (ChunMing) Zhou" <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/si.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/amdgpu/si.c +3 -3
index 40520a968eaca..28eea8317e87d 100644
@@ -1783,7 +1783,7 @@ static void si_program_aspm(struct amdgpu_device *adev)
 			if (orig != data)
 				si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_1, data);
 
-			if ((adev->family != CHIP_OLAND) && (adev->family != CHIP_HAINAN)) {
+			if ((adev->asic_type != CHIP_OLAND) && (adev->asic_type != CHIP_HAINAN)) {
 				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_0);
 				data &= ~PLL_RAMP_UP_TIME_0_MASK;
 				if (orig != data)
@@ -1832,14 +1832,14 @@ static void si_program_aspm(struct amdgpu_device *adev)
 
 			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_CNTL);
 			data &= ~LS2_EXIT_TIME_MASK;
-			if ((adev->family == CHIP_OLAND) || (adev->family == CHIP_HAINAN))
+			if ((adev->asic_type == CHIP_OLAND) || (adev->asic_type == CHIP_HAINAN))
 				data |= LS2_EXIT_TIME(5);
 			if (orig != data)
 				si_pif_phy0_wreg(adev,PB0_PIF_CNTL, data);
 
 			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_CNTL);
 			data &= ~LS2_EXIT_TIME_MASK;
-			if ((adev->family == CHIP_OLAND) || (adev->family == CHIP_HAINAN))
+			if ((adev->asic_type == CHIP_OLAND) || (adev->asic_type == CHIP_HAINAN))
 				data |= LS2_EXIT_TIME(5);
 			if (orig != data)
 				si_pif_phy1_wreg(adev,PB1_PIF_CNTL, data);
Signature · 1 line
2.20.1

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/28 drm/bridge: tc358767: Increase AUX transfer length limit

From: Andrey Smirnov <[email protected]>

[ Upstream commit e0655feaec62d5139b6b13a7b1bbb1ab8f1c2d83 ]

According to the datasheet tc358767 can transfer up to 16 bytes via its AUX channel, so the artificial limit of 8 appears to be too low. However only up to 15-bytes seem to be actually supported and trying to use 16-byte transfers results in transfers failing sporadically (with bogus status in case of I2C transfers), so limit it to 15.

---
 drivers/gpu/drm/bridge/tc358767.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/bridge/tc358767.c +1 -1
index 9705ca197b90d..cefa2c1685ba4 100644
@@ -300,7 +300,7 @@ static ssize_t tc_aux_transfer(struct drm_dp_aux *aux,
 			       struct drm_dp_aux_msg *msg)
 {
 	struct tc_data *tc = aux_to_tc(aux);
-	size_t size = min_t(size_t, 8, msg->size);
+	size_t size = min_t(size_t, DP_AUX_MAX_PAYLOAD_BYTES - 1, msg->size);
 	u8 request = msg->request & ~DP_AUX_I2C_MOT;
 	u8 *buf = msg->buffer;
 	u32 tmp = 0;
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#1 lore raw

PATCH 2/28 drm/kms: Catch mode_object lifetime errors

From: Daniel Vetter <[email protected]>

[ Upstream commit 4f5368b5541a902f6596558b05f5c21a9770dd32 ]

Only dynamic mode objects, i.e. those which are refcounted and have a free callback, can be added while the overall drm_device is visible to userspace. All others must be added before drm_dev_register and removed after drm_dev_unregister.

Small issue around drivers still using the load/unload callbacks, we need to make sure we set dev->registered so that load/unload code in these callbacks doesn't trigger false warnings. Only a small adjustement in drm_dev_register was needed.

Motivated by some irc discussions about object ids of dynamic objects like blobs become invalid, and me going on a bit an audit spree.

---
 drivers/gpu/drm/drm_drv.c         | 4 ++--
 drivers/gpu/drm/drm_mode_object.c | 4 ++++
 2 files changed, 6 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_drv.c +2 -2
index 340440febf9a5..426365181bdc1 100644
@@ -804,14 +804,14 @@ int drm_dev_register(struct drm_device *dev, unsigned long flags)
 	if (ret)
 		goto err_minors;
 
-	dev->registered = true;
-
 	if (dev->driver->load) {
 		ret = dev->driver->load(dev, flags);
 		if (ret)
 			goto err_minors;
 	}
 
+	dev->registered = true;
+
 	if (drm_core_check_feature(dev, DRIVER_MODESET))
 		drm_modeset_register_all(dev);
 
drivers/gpu/drm/drm_mode_object.c +4 -0
index 5b692ce6a45db..639f869de171b 100644
@@ -37,6 +37,8 @@ int __drm_mode_object_add(struct drm_device *dev, struct drm_mode_object *obj,
 {
 	int ret;
 
+	WARN_ON(dev->registered && !obj_free_cb);
+
 	mutex_lock(&dev->mode_config.idr_mutex);
 	ret = idr_alloc(&dev->mode_config.crtc_idr, register_obj ? obj : NULL, 1, 0, GFP_KERNEL);
 	if (ret >= 0) {
@@ -96,6 +98,8 @@ void drm_mode_object_register(struct drm_device *dev,
 void drm_mode_object_unregister(struct drm_device *dev,
 				struct drm_mode_object *object)
 {
+	WARN_ON(dev->registered && !object->free_cb);
+
 	mutex_lock(&dev->mode_config.idr_mutex);
 	if (object->id) {
 		idr_remove(&dev->mode_config.crtc_idr, object->id);
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#2 lore raw

PATCH 3/28 drm/panel: simple: fix AUO g185han01 horizontal blanking

In reply to Sasha Levin

From: Lucas Stach <[email protected]>

[ Upstream commit f8c6bfc612b56f02e1b8fae699dff12738aaf889 ]

The horizontal blanking periods are too short, as the values are specified for a single LVDS channel. Since this panel is dual LVDS they need to be doubled. With this change the panel reaches its nominal vrefresh rate of 60Fps, instead of the 64Fps with the current wrong blanking.

Philipp Zabel added:
The datasheet specifies 960 active clocks + 40/128/160 clocks blanking on each of the two LVDS channels (min/typical/max), so doubled this is now correct.

---
 drivers/gpu/drm/panel/panel-simple.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/panel/panel-simple.c +3 -3
index 7a0fd4e4e78d5..c1daed3fe8428 100644
@@ -614,9 +614,9 @@ static const struct panel_desc auo_g133han01 = {
 static const struct display_timing auo_g185han01_timings = {
 	.pixelclock = { 120000000, 144000000, 175000000 },
 	.hactive = { 1920, 1920, 1920 },
-	.hfront_porch = { 18, 60, 74 },
-	.hback_porch = { 12, 44, 54 },
-	.hsync_len = { 10, 24, 32 },
+	.hfront_porch = { 36, 120, 148 },
+	.hback_porch = { 24, 88, 108 },
+	.hsync_len = { 20, 48, 64 },
 	.vactive = { 1080, 1080, 1080 },
 	.vfront_porch = { 6, 10, 40 },
 	.vback_porch = { 2, 5, 20 },
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#11 lore raw

PATCH 12/28 clk: sirf: Don't reference clk_init_data after registration

In reply to Sasha Levin

From: Stephen Boyd <[email protected]>

[ Upstream commit af55dadfbce35b4f4c6247244ce3e44b2e242b84 ]

A future patch is going to change semantics of clk_register() so that clk_hw::init is guaranteed to be NULL after a clk is registered. Avoid referencing this member here so that we don't run into NULL pointer exceptions.

---
 drivers/clk/sirf/clk-common.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)
drivers/clk/sirf/clk-common.c +8 -4
index 77e1e2491689b..edb7197cc4b4d 100644
@@ -298,9 +298,10 @@ static u8 dmn_clk_get_parent(struct clk_hw *hw)
 {
 	struct clk_dmn *clk = to_dmnclk(hw);
 	u32 cfg = clkc_readl(clk->regofs);
+	const char *name = clk_hw_get_name(hw);
 
 	/* parent of io domain can only be pll3 */
-	if (strcmp(hw->init->name, "io") == 0)
+	if (strcmp(name, "io") == 0)
 		return 4;
 
 	WARN_ON((cfg & (BIT(3) - 1)) > 4);
@@ -312,9 +313,10 @@ static int dmn_clk_set_parent(struct clk_hw *hw, u8 parent)
 {
 	struct clk_dmn *clk = to_dmnclk(hw);
 	u32 cfg = clkc_readl(clk->regofs);
+	const char *name = clk_hw_get_name(hw);
 
 	/* parent of io domain can only be pll3 */
-	if (strcmp(hw->init->name, "io") == 0)
+	if (strcmp(name, "io") == 0)
 		return -EINVAL;
 
 	cfg &= ~(BIT(3) - 1);
@@ -354,7 +356,8 @@ static long dmn_clk_round_rate(struct clk_hw *hw, unsigned long rate,
 {
 	unsigned long fin;
 	unsigned ratio, wait, hold;
-	unsigned bits = (strcmp(hw->init->name, "mem") == 0) ? 3 : 4;
+	const char *name = clk_hw_get_name(hw);
+	unsigned bits = (strcmp(name, "mem") == 0) ? 3 : 4;
 
 	fin = *parent_rate;
 	ratio = fin / rate;
@@ -376,7 +379,8 @@ static int dmn_clk_set_rate(struct clk_hw *hw, unsigned long rate,
 	struct clk_dmn *clk = to_dmnclk(hw);
 	unsigned long fin;
 	unsigned ratio, wait, hold, reg;
-	unsigned bits = (strcmp(hw->init->name, "mem") == 0) ? 3 : 4;
+	const char *name = clk_hw_get_name(hw);
+	unsigned bits = (strcmp(name, "mem") == 0) ? 3 : 4;
 
 	fin = parent_rate;
 	ratio = fin / rate;
Signature · 1 line
2.20.1
author maintainer
#13 lore raw

PATCH 14/28 powerpc/xmon: Check for HV mode when dumping XIVE info from OPAL

In reply to Sasha Levin

From: Cédric Le Goater <[email protected]>

[ Upstream commit c3e0dbd7f780a58c4695f1cd8fc8afde80376737 ]

Currently, the xmon 'dx' command calls OPAL to dump the XIVE state in the OPAL logs and also outputs some of the fields of the internal XIVE structures in Linux. The OPAL calls can only be done on baremetal (PowerNV) and they crash a pseries machine. Fix by checking the hypervisor feature of the CPU.

---
 arch/powerpc/xmon/xmon.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
arch/powerpc/xmon/xmon.c +9 -6
index 6b9038a3e79f0..5a739588aa505 100644
@@ -2438,13 +2438,16 @@ static void dump_pacas(void)
 static void dump_one_xive(int cpu)
 {
 	unsigned int hwid = get_hard_smp_processor_id(cpu);
+	bool hv = cpu_has_feature(CPU_FTR_HVMODE);
 
-	opal_xive_dump(XIVE_DUMP_TM_HYP, hwid);
-	opal_xive_dump(XIVE_DUMP_TM_POOL, hwid);
-	opal_xive_dump(XIVE_DUMP_TM_OS, hwid);
-	opal_xive_dump(XIVE_DUMP_TM_USER, hwid);
-	opal_xive_dump(XIVE_DUMP_VP, hwid);
-	opal_xive_dump(XIVE_DUMP_EMU_STATE, hwid);
+	if (hv) {
+		opal_xive_dump(XIVE_DUMP_TM_HYP, hwid);
+		opal_xive_dump(XIVE_DUMP_TM_POOL, hwid);
+		opal_xive_dump(XIVE_DUMP_TM_OS, hwid);
+		opal_xive_dump(XIVE_DUMP_TM_USER, hwid);
+		opal_xive_dump(XIVE_DUMP_VP, hwid);
+		opal_xive_dump(XIVE_DUMP_EMU_STATE, hwid);
+	}
 
 	if (setjmp(bus_error_jmp) != 0) {
 		catch_memory_errors = 0;
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 18/28 powerpc/pseries/mobility: use cond_resched when updating device tree

In reply to Sasha Levin

From: Nathan Lynch <[email protected]>

[ Upstream commit ccfb5bd71d3d1228090a8633800ae7cdf42a94ac ]

After a partition migration, pseries_devicetree_update() processes changes to the device tree communicated from the platform to Linux. This is a relatively heavyweight operation, with multiple device tree searches, memory allocations, and conversations with partition firmware.

There's a few levels of nested loops which are bounded only by decisions made by the platform, outside of Linux's control, and indeed we have seen RCU stalls on large systems while executing this call graph. Use cond_resched() in these loops so that the cpu is yielded when needed.

---
 arch/powerpc/platforms/pseries/mobility.c | 9 +++++++++
 1 file changed, 9 insertions(+)
arch/powerpc/platforms/pseries/mobility.c +9 -0
index 4addc552eb33d..9739a055e5f7b 100644
@@ -12,6 +12,7 @@
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/kobject.h>
+#include <linux/sched.h>
 #include <linux/smp.h>
 #include <linux/stat.h>
 #include <linux/completion.h>
@@ -208,7 +209,11 @@ static int update_dt_node(__be32 phandle, s32 scope)
 
 				prop_data += vd;
 			}
+
+			cond_resched();
 		}
+
+		cond_resched();
 	} while (rtas_rc == 1);
 
 	of_node_put(dn);
@@ -317,8 +322,12 @@ int pseries_devicetree_update(s32 scope)
 					add_dt_node(phandle, drc_index);
 					break;
 				}
+
+				cond_resched();
 			}
 		}
+
+		cond_resched();
 	} while (rc == 1);
 
 	kfree(rtas_buf);
Signature · 1 line
2.20.1
author maintainer
#19 lore raw

PATCH 20/28 vfio_pci: Restore original state on release

In reply to Sasha Levin

From: hexin <[email protected]>

[ Upstream commit 92c8026854c25093946e0d7fe536fd9eac440f06 ]

vfio_pci_enable() saves the device's initial configuration information with the intent that it is restored in vfio_pci_disable(). However, the commit referenced in Fixes: below replaced the call to __pci_reset_function_locked(), which is not wrapped in a state save and restore, with pci_try_reset_function(), which overwrites the restored device state with the current state before applying it to the device. Reinstate use of __pci_reset_function_locked() to return to the desired behavior.

Fixes 890ed578df82 ("vfio-pci: Use pci "try" reset interface")
---
 drivers/vfio/pci/vfio_pci.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)
drivers/vfio/pci/vfio_pci.c +13 -4
index 6f5cc67e343e7..15b1cd4ef5a77 100644
@@ -363,11 +363,20 @@ static void vfio_pci_disable(struct vfio_pci_device *vdev)
 	pci_write_config_word(pdev, PCI_COMMAND, PCI_COMMAND_INTX_DISABLE);
 
 	/*
-	 * Try to reset the device.  The success of this is dependent on
-	 * being able to lock the device, which is not always possible.
+	 * Try to get the locks ourselves to prevent a deadlock. The
+	 * success of this is dependent on being able to lock the device,
+	 * which is not always possible.
+	 * We can not use the "try" reset interface here, which will
+	 * overwrite the previously restored configuration information.
 	 */
-	if (vdev->reset_works && !pci_try_reset_function(pdev))
-		vdev->needs_reset = false;
+	if (vdev->reset_works && pci_cfg_access_trylock(pdev)) {
+		if (device_trylock(&pdev->dev)) {
+			if (!__pci_reset_function_locked(pdev))
+				vdev->needs_reset = false;
+			device_unlock(&pdev->dev);
+		}
+		pci_cfg_access_unlock(pdev);
+	}
 
 	pci_restore_state(pdev);
 out:
Signature · 1 line
2.20.1
author maintainer
#27 lore raw

PATCH 28/28 dma-buf/sw_sync: Synchronize signal vs syncpt free

In reply to Sasha Levin

From: Chris Wilson <[email protected]>

[ Upstream commit d3c6dd1fb30d3853c2012549affe75c930f4a2f9 ]

During release of the syncpt, we remove it from the list of syncpt and the tree, but only if it is not already been removed. However, during signaling, we first remove the syncpt from the list. So, if we concurrently free and signal the syncpt, the free may decide that it is not part of the tree and immediately free itself -- meanwhile the signaler goes on to use the now freed datastructure.

In particular, we get struck by commit 0e2f733addbf ("dma-buf: make dma_fence structure a bit smaller v2") as the cb_list is immediately clobbered by the kfree_rcu.

v2: Avoid calling into timeline_fence_release() from under the spinlock

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111381

Fixes d3862e44daa7 ("dma-buf/sw-sync: Fix locking around sync_timeline lists")

References: 0e2f733addbf ("dma-buf: make dma_fence structure a bit smaller v2")

---
 drivers/dma-buf/sw_sync.c | 16 +++++++---------
 1 file changed, 7 insertions(+), 9 deletions(-)
drivers/dma-buf/sw_sync.c +7 -9
index 24f83f9eeaedc..114b36674af42 100644
@@ -141,17 +141,14 @@ static void timeline_fence_release(struct dma_fence *fence)
 {
 	struct sync_pt *pt = dma_fence_to_sync_pt(fence);
 	struct sync_timeline *parent = dma_fence_parent(fence);
+	unsigned long flags;
 
+	spin_lock_irqsave(fence->lock, flags);
 	if (!list_empty(&pt->link)) {
-		unsigned long flags;
-
-		spin_lock_irqsave(fence->lock, flags);
-		if (!list_empty(&pt->link)) {
-			list_del(&pt->link);
-			rb_erase(&pt->node, &parent->pt_tree);
-		}
-		spin_unlock_irqrestore(fence->lock, flags);
+		list_del(&pt->link);
+		rb_erase(&pt->node, &parent->pt_tree);
 	}
+	spin_unlock_irqrestore(fence->lock, flags);
 
 	sync_timeline_put(parent);
 	dma_fence_free(fence);
@@ -275,7 +272,8 @@ static struct sync_pt *sync_pt_create(struct sync_timeline *obj,
 				p = &parent->rb_left;
 			} else {
 				if (dma_fence_get_rcu(&other->base)) {
-					dma_fence_put(&pt->base);
+					sync_timeline_put(obj);
+					kfree(pt);
 					pt = other;
 					goto unlock;
 				}
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#12 lore raw

PATCH 13/28 clk: zx296718: Don't reference clk_init_data after registration

In reply to Sasha Levin

From: Stephen Boyd <[email protected]>

[ Upstream commit 1a4549c150e27dbc3aea762e879a88209df6d1a5 ]

A future patch is going to change semantics of clk_register() so that clk_hw::init is guaranteed to be NULL after a clk is registered. Avoid referencing this member here so that we don't run into NULL pointer exceptions.

---
 drivers/clk/zte/clk-zx296718.c | 109 +++++++++++++++------------------
 1 file changed, 49 insertions(+), 60 deletions(-)
drivers/clk/zte/clk-zx296718.c +49 -60
index 354dd508c5169..8dfb8523b79db 100644
@@ -567,6 +567,7 @@ static int __init top_clocks_init(struct device_node *np)
 {
 	void __iomem *reg_base;
 	int i, ret;
+	const char *name;
 
 	reg_base = of_iomap(np, 0);
 	if (!reg_base) {
@@ -576,11 +577,10 @@ static int __init top_clocks_init(struct device_node *np)
 
 	for (i = 0; i < ARRAY_SIZE(zx296718_pll_clk); i++) {
 		zx296718_pll_clk[i].reg_base += (uintptr_t)reg_base;
+		name = zx296718_pll_clk[i].hw.init->name;
 		ret = clk_hw_register(NULL, &zx296718_pll_clk[i].hw);
-		if (ret) {
-			pr_warn("top clk %s init error!\n",
-				zx296718_pll_clk[i].hw.init->name);
-		}
+		if (ret)
+			pr_warn("top clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(top_ffactor_clk); i++) {
@@ -588,11 +588,10 @@ static int __init top_clocks_init(struct device_node *np)
 			top_hw_onecell_data.hws[top_ffactor_clk[i].id] =
 					&top_ffactor_clk[i].factor.hw;
 
+		name = top_ffactor_clk[i].factor.hw.init->name;
 		ret = clk_hw_register(NULL, &top_ffactor_clk[i].factor.hw);
-		if (ret) {
-			pr_warn("top clk %s init error!\n",
-				top_ffactor_clk[i].factor.hw.init->name);
-		}
+		if (ret)
+			pr_warn("top clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(top_mux_clk); i++) {
@@ -601,11 +600,10 @@ static int __init top_clocks_init(struct device_node *np)
 					&top_mux_clk[i].mux.hw;
 
 		top_mux_clk[i].mux.reg += (uintptr_t)reg_base;
+		name = top_mux_clk[i].mux.hw.init->name;
 		ret = clk_hw_register(NULL, &top_mux_clk[i].mux.hw);
-		if (ret) {
-			pr_warn("top clk %s init error!\n",
-				top_mux_clk[i].mux.hw.init->name);
-		}
+		if (ret)
+			pr_warn("top clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(top_gate_clk); i++) {
@@ -614,11 +612,10 @@ static int __init top_clocks_init(struct device_node *np)
 					&top_gate_clk[i].gate.hw;
 
 		top_gate_clk[i].gate.reg += (uintptr_t)reg_base;
+		name = top_gate_clk[i].gate.hw.init->name;
 		ret = clk_hw_register(NULL, &top_gate_clk[i].gate.hw);
-		if (ret) {
-			pr_warn("top clk %s init error!\n",
-				top_gate_clk[i].gate.hw.init->name);
-		}
+		if (ret)
+			pr_warn("top clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(top_div_clk); i++) {
@@ -627,11 +624,10 @@ static int __init top_clocks_init(struct device_node *np)
 					&top_div_clk[i].div.hw;
 
 		top_div_clk[i].div.reg += (uintptr_t)reg_base;
+		name = top_div_clk[i].div.hw.init->name;
 		ret = clk_hw_register(NULL, &top_div_clk[i].div.hw);
-		if (ret) {
-			pr_warn("top clk %s init error!\n",
-				top_div_clk[i].div.hw.init->name);
-		}
+		if (ret)
+			pr_warn("top clk %s init error!\n", name);
 	}
 
 	ret = of_clk_add_hw_provider(np, of_clk_hw_onecell_get,
@@ -757,6 +753,7 @@ static int __init lsp0_clocks_init(struct device_node *np)
 {
 	void __iomem *reg_base;
 	int i, ret;
+	const char *name;
 
 	reg_base = of_iomap(np, 0);
 	if (!reg_base) {
@@ -770,11 +767,10 @@ static int __init lsp0_clocks_init(struct device_node *np)
 					&lsp0_mux_clk[i].mux.hw;
 
 		lsp0_mux_clk[i].mux.reg += (uintptr_t)reg_base;
+		name = lsp0_mux_clk[i].mux.hw.init->name;
 		ret = clk_hw_register(NULL, &lsp0_mux_clk[i].mux.hw);
-		if (ret) {
-			pr_warn("lsp0 clk %s init error!\n",
-				lsp0_mux_clk[i].mux.hw.init->name);
-		}
+		if (ret)
+			pr_warn("lsp0 clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(lsp0_gate_clk); i++) {
@@ -783,11 +779,10 @@ static int __init lsp0_clocks_init(struct device_node *np)
 					&lsp0_gate_clk[i].gate.hw;
 
 		lsp0_gate_clk[i].gate.reg += (uintptr_t)reg_base;
+		name = lsp0_gate_clk[i].gate.hw.init->name;
 		ret = clk_hw_register(NULL, &lsp0_gate_clk[i].gate.hw);
-		if (ret) {
-			pr_warn("lsp0 clk %s init error!\n",
-				lsp0_gate_clk[i].gate.hw.init->name);
-		}
+		if (ret)
+			pr_warn("lsp0 clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(lsp0_div_clk); i++) {
@@ -796,11 +791,10 @@ static int __init lsp0_clocks_init(struct device_node *np)
 					&lsp0_div_clk[i].div.hw;
 
 		lsp0_div_clk[i].div.reg += (uintptr_t)reg_base;
+		name = lsp0_div_clk[i].div.hw.init->name;
 		ret = clk_hw_register(NULL, &lsp0_div_clk[i].div.hw);
-		if (ret) {
-			pr_warn("lsp0 clk %s init error!\n",
-				lsp0_div_clk[i].div.hw.init->name);
-		}
+		if (ret)
+			pr_warn("lsp0 clk %s init error!\n", name);
 	}
 
 	ret = of_clk_add_hw_provider(np, of_clk_hw_onecell_get,
@@ -865,6 +859,7 @@ static int __init lsp1_clocks_init(struct device_node *np)
 {
 	void __iomem *reg_base;
 	int i, ret;
+	const char *name;
 
 	reg_base = of_iomap(np, 0);
 	if (!reg_base) {
@@ -878,11 +873,10 @@ static int __init lsp1_clocks_init(struct device_node *np)
 					&lsp0_mux_clk[i].mux.hw;
 
 		lsp1_mux_clk[i].mux.reg += (uintptr_t)reg_base;
+		name = lsp1_mux_clk[i].mux.hw.init->name;
 		ret = clk_hw_register(NULL, &lsp1_mux_clk[i].mux.hw);
-		if (ret) {
-			pr_warn("lsp1 clk %s init error!\n",
-				lsp1_mux_clk[i].mux.hw.init->name);
-		}
+		if (ret)
+			pr_warn("lsp1 clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(lsp1_gate_clk); i++) {
@@ -891,11 +885,10 @@ static int __init lsp1_clocks_init(struct device_node *np)
 					&lsp1_gate_clk[i].gate.hw;
 
 		lsp1_gate_clk[i].gate.reg += (uintptr_t)reg_base;
+		name = lsp1_gate_clk[i].gate.hw.init->name;
 		ret = clk_hw_register(NULL, &lsp1_gate_clk[i].gate.hw);
-		if (ret) {
-			pr_warn("lsp1 clk %s init error!\n",
-				lsp1_gate_clk[i].gate.hw.init->name);
-		}
+		if (ret)
+			pr_warn("lsp1 clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(lsp1_div_clk); i++) {
@@ -904,11 +897,10 @@ static int __init lsp1_clocks_init(struct device_node *np)
 					&lsp1_div_clk[i].div.hw;
 
 		lsp1_div_clk[i].div.reg += (uintptr_t)reg_base;
+		name = lsp1_div_clk[i].div.hw.init->name;
 		ret = clk_hw_register(NULL, &lsp1_div_clk[i].div.hw);
-		if (ret) {
-			pr_warn("lsp1 clk %s init error!\n",
-				lsp1_div_clk[i].div.hw.init->name);
-		}
+		if (ret)
+			pr_warn("lsp1 clk %s init error!\n", name);
 	}
 
 	ret = of_clk_add_hw_provider(np, of_clk_hw_onecell_get,
@@ -982,6 +974,7 @@ static int __init audio_clocks_init(struct device_node *np)
 {
 	void __iomem *reg_base;
 	int i, ret;
+	const char *name;
 
 	reg_base = of_iomap(np, 0);
 	if (!reg_base) {
@@ -995,11 +988,10 @@ static int __init audio_clocks_init(struct device_node *np)
 					&audio_mux_clk[i].mux.hw;
 
 		audio_mux_clk[i].mux.reg += (uintptr_t)reg_base;
+		name = audio_mux_clk[i].mux.hw.init->name;
 		ret = clk_hw_register(NULL, &audio_mux_clk[i].mux.hw);
-		if (ret) {
-			pr_warn("audio clk %s init error!\n",
-				audio_mux_clk[i].mux.hw.init->name);
-		}
+		if (ret)
+			pr_warn("audio clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(audio_adiv_clk); i++) {
@@ -1008,11 +1000,10 @@ static int __init audio_clocks_init(struct device_node *np)
 					&audio_adiv_clk[i].hw;
 
 		audio_adiv_clk[i].reg_base += (uintptr_t)reg_base;
+		name = audio_adiv_clk[i].hw.init->name;
 		ret = clk_hw_register(NULL, &audio_adiv_clk[i].hw);
-		if (ret) {
-			pr_warn("audio clk %s init error!\n",
-				audio_adiv_clk[i].hw.init->name);
-		}
+		if (ret)
+			pr_warn("audio clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(audio_div_clk); i++) {
@@ -1021,11 +1012,10 @@ static int __init audio_clocks_init(struct device_node *np)
 					&audio_div_clk[i].div.hw;
 
 		audio_div_clk[i].div.reg += (uintptr_t)reg_base;
+		name = audio_div_clk[i].div.hw.init->name;
 		ret = clk_hw_register(NULL, &audio_div_clk[i].div.hw);
-		if (ret) {
-			pr_warn("audio clk %s init error!\n",
-				audio_div_clk[i].div.hw.init->name);
-		}
+		if (ret)
+			pr_warn("audio clk %s init error!\n", name);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(audio_gate_clk); i++) {
@@ -1034,11 +1024,10 @@ static int __init audio_clocks_init(struct device_node *np)
 					&audio_gate_clk[i].gate.hw;
 
 		audio_gate_clk[i].gate.reg += (uintptr_t)reg_base;
+		name = audio_gate_clk[i].gate.hw.init->name;
 		ret = clk_hw_register(NULL, &audio_gate_clk[i].gate.hw);
-		if (ret) {
-			pr_warn("audio clk %s init error!\n",
-				audio_gate_clk[i].gate.hw.init->name);
-		}
+		if (ret)
+			pr_warn("audio clk %s init error!\n", name);
 	}
 
 	ret = of_clk_add_hw_provider(np, of_clk_hw_onecell_get,
Signature · 1 line
2.20.1
author maintainer
#22 lore raw

PATCH 23/28 powerpc/64s/exception: machine check use correct cfar for late handler

In reply to Sasha Levin

From: Nicholas Piggin <[email protected]>

[ Upstream commit 0b66370c61fcf5fcc1d6901013e110284da6e2bb ]

Bare metal machine checks run an "early" handler in real mode before running the main handler which reports the event.

The main handler runs exactly as a normal interrupt handler, after the "windup" which sets registers back as they were at interrupt entry. CFAR does not get restored by the windup code, so that will be wrong when the handler is run.

Restore the CFAR to the saved value before running the late handler.

---
 arch/powerpc/kernel/exceptions-64s.S | 4 ++++
 1 file changed, 4 insertions(+)
arch/powerpc/kernel/exceptions-64s.S +4 -0
index 43cde6c602795..cdc53fd905977 100644
@@ -464,6 +464,10 @@ EXC_COMMON_BEGIN(machine_check_handle_early)
 	RFI_TO_USER_OR_KERNEL
 9:
 	/* Deliver the machine check to host kernel in V mode. */
+BEGIN_FTR_SECTION
+	ld	r10,ORIG_GPR3(r1)
+	mtspr	SPRN_CFAR,r10
+END_FTR_SECTION_IFSET(CPU_FTR_CFAR)
 	MACHINE_CHECK_HANDLER_WINDUP
 	b	machine_check_pSeries
 
Signature · 1 line
2.20.1
author maintainer
#26 lore raw

PATCH 27/28 scsi: core: Reduce memory required for SCSI logging

In reply to Sasha Levin

From: Bart Van Assche <[email protected]>

[ Upstream commit dccc96abfb21dc19d69e707c38c8ba439bba7160 ]

The data structure used for log messages is so large that it can cause a boot failure. Since allocations from that data structure can fail anyway, use kmalloc() / kfree() instead of that data structure.

See also https://bugzilla.kernel.org/show_bug.cgi?id=204119.
See also commit ded85c193a39 ("scsi: Implement per-cpu logging buffer") # v4.0.

---
 drivers/scsi/scsi_logging.c | 48 +++----------------------------------
 include/scsi/scsi_dbg.h     |  2 --
 2 files changed, 3 insertions(+), 47 deletions(-)
drivers/scsi/scsi_logging.c +3 -45
index bd70339c1242e..03d9855a6afd7 100644
@@ -16,57 +16,15 @@
 #include <scsi/scsi_eh.h>
 #include <scsi/scsi_dbg.h>
 
-#define SCSI_LOG_SPOOLSIZE 4096
-
-#if (SCSI_LOG_SPOOLSIZE / SCSI_LOG_BUFSIZE) > BITS_PER_LONG
-#warning SCSI logging bitmask too large
-#endif
-
-struct scsi_log_buf {
-	char buffer[SCSI_LOG_SPOOLSIZE];
-	unsigned long map;
-};
-
-static DEFINE_PER_CPU(struct scsi_log_buf, scsi_format_log);
-
 static char *scsi_log_reserve_buffer(size_t *len)
 {
-	struct scsi_log_buf *buf;
-	unsigned long map_bits = sizeof(buf->buffer) / SCSI_LOG_BUFSIZE;
-	unsigned long idx = 0;
-
-	preempt_disable();
-	buf = this_cpu_ptr(&scsi_format_log);
-	idx = find_first_zero_bit(&buf->map, map_bits);
-	if (likely(idx < map_bits)) {
-		while (test_and_set_bit(idx, &buf->map)) {
-			idx = find_next_zero_bit(&buf->map, map_bits, idx);
-			if (idx >= map_bits)
-				break;
-		}
-	}
-	if (WARN_ON(idx >= map_bits)) {
-		preempt_enable();
-		return NULL;
-	}
-	*len = SCSI_LOG_BUFSIZE;
-	return buf->buffer + idx * SCSI_LOG_BUFSIZE;
+	*len = 128;
+	return kmalloc(*len, GFP_ATOMIC);
 }
 
 static void scsi_log_release_buffer(char *bufptr)
 {
-	struct scsi_log_buf *buf;
-	unsigned long idx;
-	int ret;
-
-	buf = this_cpu_ptr(&scsi_format_log);
-	if (bufptr >= buf->buffer &&
-	    bufptr < buf->buffer + SCSI_LOG_SPOOLSIZE) {
-		idx = (bufptr - buf->buffer) / SCSI_LOG_BUFSIZE;
-		ret = test_and_clear_bit(idx, &buf->map);
-		WARN_ON(!ret);
-	}
-	preempt_enable();
+	kfree(bufptr);
 }
 
 static inline const char *scmd_name(const struct scsi_cmnd *scmd)
include/scsi/scsi_dbg.h +0 -2
index 04e0679767f63..2b5dfae782722 100644
@@ -6,8 +6,6 @@ struct scsi_cmnd;
 struct scsi_device;
 struct scsi_sense_hdr;
 
-#define SCSI_LOG_BUFSIZE 128
-
 extern void scsi_print_command(struct scsi_cmnd *);
 extern size_t __scsi_format_command(char *, size_t,
 				   const unsigned char *, size_t);
Signature · 1 line
2.20.1
author maintainer
#25 lore raw

PATCH 26/28 clk: at91: select parent if main oscillator or bypass is enabled

In reply to Sasha Levin

From: Eugen Hristev <[email protected]>

[ Upstream commit 69a6bcde7fd3fe6f3268ce26f31d9d9378384c98 ]

Selecting the right parent for the main clock is done using only main oscillator enabled bit.
In case we have this oscillator bypassed by an external signal (no driving on the XOUT line), we still use external clock, but with BYPASS bit set. So, in this case we must select the same parent as before. Create a macro that will select the right parent considering both bits from the MOR register.
Use this macro when looking for the right parent.

---
 drivers/clk/at91/clk-main.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
drivers/clk/at91/clk-main.c +7 -3
index c813c27f2e58c..2f97a843d6d6b 100644
@@ -27,6 +27,10 @@
 
 #define MOR_KEY_MASK		(0xff << 16)
 
+#define clk_main_parent_select(s)	(((s) & \
+					(AT91_PMC_MOSCEN | \
+					AT91_PMC_OSCBYPASS)) ? 1 : 0)
+
 struct clk_main_osc {
 	struct clk_hw hw;
 	struct regmap *regmap;
@@ -119,7 +123,7 @@ static int clk_main_osc_is_prepared(struct clk_hw *hw)
 
 	regmap_read(regmap, AT91_PMC_SR, &status);
 
-	return (status & AT91_PMC_MOSCS) && (tmp & AT91_PMC_MOSCEN);
+	return (status & AT91_PMC_MOSCS) && clk_main_parent_select(tmp);
 }
 
 static const struct clk_ops main_osc_ops = {
@@ -530,7 +534,7 @@ static u8 clk_sam9x5_main_get_parent(struct clk_hw *hw)
 
 	regmap_read(clkmain->regmap, AT91_CKGR_MOR, &status);
 
-	return status & AT91_PMC_MOSCEN ? 1 : 0;
+	return clk_main_parent_select(status);
 }
 
 static const struct clk_ops sam9x5_main_ops = {
@@ -572,7 +576,7 @@ at91_clk_register_sam9x5_main(struct regmap *regmap,
 	clkmain->hw.init = &init;
 	clkmain->regmap = regmap;
 	regmap_read(clkmain->regmap, AT91_CKGR_MOR, &status);
-	clkmain->parent = status & AT91_PMC_MOSCEN ? 1 : 0;
+	clkmain->parent = clk_main_parent_select(status);
 
 	hw = &clkmain->hw;
 	ret = clk_hw_register(NULL, &clkmain->hw);
Signature · 1 line
2.20.1
author maintainer
#14 lore raw

PATCH 15/28 powerpc/rtas: use device model APIs and serialization during LPM

In reply to Sasha Levin

From: Nathan Lynch <[email protected]>

[ Upstream commit a6717c01ddc259f6f73364779df058e2c67309f8 ]

The LPAR migration implementation and userspace-initiated cpu hotplug can interleave their executions like so:

1. Set cpu 7 offline via sysfs.

2. Begin a partition migration, whose implementation requires the OS to ensure all present cpus are online; cpu 7 is onlined:

     rtas_ibm_suspend_me -> rtas_online_cpus_mask -> cpu_up

This sets cpu 7 online in all respects except for the cpu's corresponding struct device; dev->offline remains true.

3. Set cpu 7 online via sysfs. _cpu_up() determines that cpu 7 is already online and returns success. The driver core (device_online) sets dev->offline = false.

4. The migration completes and restores cpu 7 to offline state:

     rtas_ibm_suspend_me -> rtas_offline_cpus_mask -> cpu_down

This leaves cpu7 in a state where the driver core considers the cpu device online, but in all other respects it is offline and unused. Attempts to online the cpu via sysfs appear to succeed but the driver core actually does not pass the request to the lower-level cpuhp support code. This makes the cpu unusable until the cpu device is manually set offline and then online again via sysfs.

Instead of directly calling cpu_up/cpu_down, the migration code should use the higher-level device core APIs to maintain consistent state and serialize operations.

Fixes 120496ac2d2d ("powerpc: Bring all threads online prior to migration/hibernation")
---
 arch/powerpc/kernel/rtas.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
arch/powerpc/kernel/rtas.c +8 -3
index 1643e9e536557..141d192c69538 100644
@@ -874,15 +874,17 @@ static int rtas_cpu_state_change_mask(enum rtas_cpu_state state,
 		return 0;
 
 	for_each_cpu(cpu, cpus) {
+		struct device *dev = get_cpu_device(cpu);
+
 		switch (state) {
 		case DOWN:
-			cpuret = cpu_down(cpu);
+			cpuret = device_offline(dev);
 			break;
 		case UP:
-			cpuret = cpu_up(cpu);
+			cpuret = device_online(dev);
 			break;
 		}
-		if (cpuret) {
+		if (cpuret < 0) {
 			pr_debug("%s: cpu_%s for cpu#%d returned %d.\n",
 					__func__,
 					((state == UP) ? "up" : "down"),
@@ -971,6 +973,8 @@ int rtas_ibm_suspend_me(u64 handle)
 	data.token = rtas_token("ibm,suspend-me");
 	data.complete = &done;
 
+	lock_device_hotplug();
+
 	/* All present CPUs must be online */
 	cpumask_andnot(offline_mask, cpu_present_mask, cpu_online_mask);
 	cpuret = rtas_online_cpus_mask(offline_mask);
@@ -1002,6 +1006,7 @@ int rtas_ibm_suspend_me(u64 handle)
 				__func__);
 
 out:
+	unlock_device_hotplug();
 	free_cpumask_var(offline_mask);
 	return atomic_read(&data.error);
 }
Signature · 1 line
2.20.1
author maintainer
#23 lore raw

PATCH 24/28 powerpc/pseries: correctly track irq state in default idle

In reply to Sasha Levin

From: Nathan Lynch <[email protected]>

[ Upstream commit 92c94dfb69e350471473fd3075c74bc68150879e ]

prep_irq_for_idle() is intended to be called before entering H_CEDE (and it is used by the pseries cpuidle driver). However the default pseries idle routine does not call it, leading to mismanaged lazy irq state when the cpuidle driver isn't in use. Manifestations of this include:

* Dropped IPIs in the time immediately after a cpu comes online (before it has installed the cpuidle handler), making the online operation block indefinitely waiting for the new cpu to respond.

* Hitting this WARN_ON in arch_local_irq_restore():

	/*
	 * We should already be hard disabled here. We had bugs
	 * where that wasn't the case so let's dbl check it and
	 * warn if we are wrong. Only do that when IRQ tracing
	 * is enabled as mfmsr() can be costly.
	 */
	if (WARN_ON_ONCE(mfmsr() & MSR_EE))
		__hard_irq_disable();

Call prep_irq_for_idle() from pseries_lpar_idle() and honor its result.

Fixes 363edbe2614a ("powerpc: Default arch idle could cede processor on pseries")
---
 arch/powerpc/platforms/pseries/setup.c | 3 +++
 1 file changed, 3 insertions(+)
arch/powerpc/platforms/pseries/setup.c +3 -0
index 6a0ad56e89b93..7a9945b350536 100644
@@ -307,6 +307,9 @@ static void pseries_lpar_idle(void)
 	 * low power mode by ceding processor to hypervisor
 	 */
 
+	if (!prep_irq_for_idle())
+		return;
+
 	/* Indicate to hypervisor that we are idle. */
 	get_lppaca()->idle = 1;
 
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 9/28 clk: qoriq: Fix -Wunused-const-variable

In reply to Sasha Levin

From: Nathan Huckleberry <[email protected]>

[ Upstream commit a95fb581b144b5e73da382eaedb2e32027610597 ]

drivers/clk/clk-qoriq.c:138:38: warning: unused variable 'p5020_cmux_grp1' [-Wunused-const-variable] static const struct clockgen_muxinfo p5020_cmux_grp1

drivers/clk/clk-qoriq.c:146:38: warning: unused variable 'p5020_cmux_grp2' [-Wunused-const-variable] static const struct clockgen_muxinfo p5020_cmux_grp2

In the definition of the p5020 chip, the p2041 chip's info was used instead. The p5020 and p2041 chips have different info. This is most likely a typo.

---
 drivers/clk/clk-qoriq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/clk/clk-qoriq.c +1 -1
index b0ea753b8709d..1a292519d84f2 100644
@@ -610,7 +610,7 @@ static const struct clockgen_chipinfo chipinfo[] = {
 		.guts_compat = "fsl,qoriq-device-config-1.0",
 		.init_periph = p5020_init_periph,
 		.cmux_groups = {
-			&p2041_cmux_grp1, &p2041_cmux_grp2
+			&p5020_cmux_grp1, &p5020_cmux_grp2
 		},
 		.cmux_to_group = {
 			0, 1, -1
Signature · 1 line
2.20.1
author maintainer
#24 lore raw

PATCH 25/28 arm64: fix unreachable code issue with cmpxchg

In reply to Sasha Levin

From: Arnd Bergmann <[email protected]>

[ Upstream commit 920fdab7b3ce98c14c840261e364f490f3679a62 ]

On arm64 build with clang, sometimes the __cmpxchg_mb is not inlined when CONFIG_OPTIMIZE_INLINING is set.
Clang then fails a compile-time assertion, because it cannot tell at compile time what the size of the argument is:

mm/memcontrol.o: In function `__cmpxchg_mb':
memcontrol.c:(.text+0x1a4c): undefined reference to __compiletime_assert_175' memcontrol.c:(.text+0x1a4c): relocation truncated to fit: R_AARCH64_CALL26 against undefined symbol __compiletime_assert_175'

Mark all of the cmpxchg() style functions as __always_inline to ensure that the compiler can see the result.

---
 arch/arm64/include/asm/cmpxchg.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
arch/arm64/include/asm/cmpxchg.h +3 -3
index 0f2e1ab5e1666..9b2e2e2e728ae 100644
@@ -73,7 +73,7 @@ __XCHG_CASE( ,  ,  mb_8, dmb ish, nop,  , a, l, "memory")
 #undef __XCHG_CASE
 
 #define __XCHG_GEN(sfx)							\
-static inline unsigned long __xchg##sfx(unsigned long x,		\
+static __always_inline  unsigned long __xchg##sfx(unsigned long x,	\
 					volatile void *ptr,		\
 					int size)			\
 {									\
@@ -115,7 +115,7 @@ __XCHG_GEN(_mb)
 #define xchg(...)		__xchg_wrapper( _mb, __VA_ARGS__)
 
 #define __CMPXCHG_GEN(sfx)						\
-static inline unsigned long __cmpxchg##sfx(volatile void *ptr,		\
+static __always_inline unsigned long __cmpxchg##sfx(volatile void *ptr,	\
 					   unsigned long old,		\
 					   unsigned long new,		\
 					   int size)			\
@@ -248,7 +248,7 @@ __CMPWAIT_CASE( ,  , 8);
 #undef __CMPWAIT_CASE
 
 #define __CMPWAIT_GEN(sfx)						\
-static inline void __cmpwait##sfx(volatile void *ptr,			\
+static __always_inline void __cmpwait##sfx(volatile void *ptr,		\
 				  unsigned long val,			\
 				  int size)				\
 {									\
Signature · 1 line
2.20.1
author maintainer
#15 lore raw

PATCH 16/28 powerpc/futex: Fix warning: 'oldval' may be used uninitialized in this function

In reply to Sasha Levin

From: Christophe Leroy <[email protected]>

[ Upstream commit 38a0d0cdb46d3f91534e5b9839ec2d67be14c59d ]

We see warnings such as:
kernel/futex.c: In function 'do_futex':
kernel/futex.c:1676:17: warning: 'oldval' may be used uninitialized in this function [-Wmaybe-uninitialized]

     return oldval == cmparg;
                   ^
  kernel/futex.c:1651:6: note: 'oldval' was declared here
    int oldval, ret;
        ^

This is because arch_futex_atomic_op_inuser() only sets *oval if ret is 0 and GCC doesn't see that it will only use it when ret is 0.

Anyway, the non-zero ret path is an error path that won't suffer from setting *oval, and as *oval is a local var in futex_atomic_op_inuser() it will have no impact.

---
 arch/powerpc/include/asm/futex.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
arch/powerpc/include/asm/futex.h +1 -2
index 1a944c18c5390..3c7d859452294 100644
@@ -59,8 +59,7 @@ static inline int arch_futex_atomic_op_inuser(int op, int oparg, int *oval,
 
 	pagefault_enable();
 
-	if (!ret)
-		*oval = oldval;
+	*oval = oldval;
 
 	return ret;
 }
Signature · 1 line
2.20.1
author maintainer
#16 lore raw

PATCH 17/28 powerpc/64s/radix: Remove redundant pfn_pte bitop, add VM_BUG_ON

In reply to Sasha Levin

From: Nicholas Piggin <[email protected]>

[ Upstream commit 6bb25170d7a44ef0ed9677814600f0785e7421d1 ]

pfn_pte is never given a pte above the addressable physical memory limit, so the masking is redundant. In case of a software bug, it is not obviously better to silently truncate the pfn than to corrupt the pte (either one will result in memory corruption or crashes), so there is no reason to add this to the fast path.

Add VM_BUG_ON to catch cases where the pfn is invalid. These would catch the create_section_mapping bug fixed by a previous commit.

  [16885.256466] ------------[ cut here ]------------
  [16885.256492] kernel BUG at arch/powerpc/include/asm/book3s/64/pgtable.h:612!
  cpu 0x0: Vector: 700 (Program Check) at [c0000000ee0a36d0]
      pc: c000000000080738: __map_kernel_page+0x248/0x6f0
      lr: c000000000080ac0: __map_kernel_page+0x5d0/0x6f0
      sp: c0000000ee0a3960
     msr: 9000000000029033
    current = 0xc0000000ec63b400
    paca    = 0xc0000000017f0000   irqmask: 0x03   irq_happened: 0x01
      pid   = 85, comm = sh
  kernel BUG at arch/powerpc/include/asm/book3s/64/pgtable.h:612!
  Linux version 5.3.0-rc1-00001-g0fe93e5f3394
  enter ? for help
  [c0000000ee0a3a00] c000000000d37378 create_physical_mapping+0x260/0x360
  [c0000000ee0a3b10] c000000000d370bc create_section_mapping+0x1c/0x3c
  [c0000000ee0a3b30] c000000000071f54 arch_add_memory+0x74/0x130
---
 arch/powerpc/include/asm/book3s/64/pgtable.h | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/powerpc/include/asm/book3s/64/pgtable.h +4 -2
index 4dd13b503dbbd..5f37d8c3b4798 100644
@@ -555,8 +555,10 @@ static inline int pte_present(pte_t pte)
  */
 static inline pte_t pfn_pte(unsigned long pfn, pgprot_t pgprot)
 {
-	return __pte((((pte_basic_t)(pfn) << PAGE_SHIFT) & PTE_RPN_MASK) |
-		     pgprot_val(pgprot));
+	VM_BUG_ON(pfn >> (64 - PAGE_SHIFT));
+	VM_BUG_ON((pfn << PAGE_SHIFT) & ~PTE_RPN_MASK);
+
+	return __pte(((pte_basic_t)pfn << PAGE_SHIFT) | pgprot_val(pgprot));
 }
 
 static inline unsigned long pte_pfn(pte_t pte)
Signature · 1 line
2.20.1
author maintainer
#9 lore raw

PATCH 10/28 clk: sunxi-ng: v3s: add missing clock slices for MMC2 module clocks

In reply to Sasha Levin

From: Icenowy Zheng <[email protected]>

[ Upstream commit 720099603d1f62e37b789366d7e89824b009ca28 ]

The MMC2 clock slices are currently not defined in V3s CCU driver, which makes MMC2 not working.

Fix this issue.

Fixes d0f11d14b0bc ("clk: sunxi-ng: add support for V3s CCU")
---
 drivers/clk/sunxi-ng/ccu-sun8i-v3s.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/clk/sunxi-ng/ccu-sun8i-v3s.c +3 -0
index ac12f261f8caa..9e3f4088724b4 100644
@@ -499,6 +499,9 @@ static struct clk_hw_onecell_data sun8i_v3s_hw_clks = {
 		[CLK_MMC1]		= &mmc1_clk.common.hw,
 		[CLK_MMC1_SAMPLE]	= &mmc1_sample_clk.common.hw,
 		[CLK_MMC1_OUTPUT]	= &mmc1_output_clk.common.hw,
+		[CLK_MMC2]		= &mmc2_clk.common.hw,
+		[CLK_MMC2_SAMPLE]	= &mmc2_sample_clk.common.hw,
+		[CLK_MMC2_OUTPUT]	= &mmc2_output_clk.common.hw,
 		[CLK_CE]		= &ce_clk.common.hw,
 		[CLK_SPI0]		= &spi0_clk.common.hw,
 		[CLK_USB_PHY0]		= &usb_phy0_clk.common.hw,
Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 8/28 ipmi_si: Only schedule continuously in the thread in maintenance mode

In reply to Sasha Levin

From: Corey Minyard <[email protected]>

[ Upstream commit 340ff31ab00bca5c15915e70ad9ada3030c98cf8 ]

ipmi_thread() uses back-to-back schedule() to poll for command completion which, on some machines, can push up CPU consumption and heavily tax the scheduler locks leading to noticeable overall performance degradation.

This was originally added so firmware updates through IPMI would complete in a timely manner. But we can't kill the scheduler locks for that one use case.

Instead, only run schedule() continuously in maintenance mode, where firmware updates should run.

---
 drivers/char/ipmi/ipmi_si_intf.c | 24 +++++++++++++++++++-----
 1 file changed, 19 insertions(+), 5 deletions(-)
drivers/char/ipmi/ipmi_si_intf.c +19 -5
index a106cf7b5ee02..f6ba90b90503f 100644
@@ -284,6 +284,9 @@ struct smi_info {
 	 */
 	bool irq_enable_broken;
 
+	/* Is the driver in maintenance mode? */
+	bool in_maintenance_mode;
+
 	/*
 	 * Did we get an attention that we did not handle?
 	 */
@@ -1094,11 +1097,20 @@ static int ipmi_thread(void *data)
 		spin_unlock_irqrestore(&(smi_info->si_lock), flags);
 		busy_wait = ipmi_thread_busy_wait(smi_result, smi_info,
 						  &busy_until);
-		if (smi_result == SI_SM_CALL_WITHOUT_DELAY)
+		if (smi_result == SI_SM_CALL_WITHOUT_DELAY) {
 			; /* do nothing */
-		else if (smi_result == SI_SM_CALL_WITH_DELAY && busy_wait)
-			schedule();
-		else if (smi_result == SI_SM_IDLE) {
+		} else if (smi_result == SI_SM_CALL_WITH_DELAY && busy_wait) {
+			/*
+			 * In maintenance mode we run as fast as
+			 * possible to allow firmware updates to
+			 * complete as fast as possible, but normally
+			 * don't bang on the scheduler.
+			 */
+			if (smi_info->in_maintenance_mode)
+				schedule();
+			else
+				usleep_range(100, 200);
+		} else if (smi_result == SI_SM_IDLE) {
 			if (atomic_read(&smi_info->need_watch)) {
 				schedule_timeout_interruptible(100);
 			} else {
@@ -1106,8 +1118,9 @@ static int ipmi_thread(void *data)
 				__set_current_state(TASK_INTERRUPTIBLE);
 				schedule();
 			}
-		} else
+		} else {
 			schedule_timeout_interruptible(1);
+		}
 	}
 	return 0;
 }
@@ -1286,6 +1299,7 @@ static void set_maintenance_mode(void *send_info, bool enable)
 
 	if (!enable)
 		atomic_set(&smi_info->req_events, 0);
+	smi_info->in_maintenance_mode = enable;
 }
 
 static const struct ipmi_smi_handlers handlers = {
Signature · 1 line
2.20.1