Skip to contentSkip to conversation

staging: gasket: sundry fixes and fixups

PATCH Superseded Todd Poynor 4 participants 52 messages Jul 14, 2018 – Jul 17, 2018 1 message not in the archive Staging

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

cover staging: gasket: sundry fixes and fixups

From: Todd Poynor <[email protected]>

Various fixes mainly from the chromium review of the gasket and apex drivers. More to come.

Various (18):
staging: gasket: remove X86 Kconfig restriction
MAINTAINERS: Add maintainer for drivers/staging/gasket
staging: gasket: typo and whitespace cleanups
staging: gasket: device registration error and unregister fixups
staging: gasket: sysfs mapping creation fixups
staging: gasket: fix deadlock in pci driver unregister path
staging: gasket: convert gasket_mmap_has_permissions to bool return
staging: gasket: gasket_wait_with_reschedule fixups
staging: gasket: bail out of reset sequence on device callback error
staging: gasket: gasket_open use container_of()
staging: gasket: always allow root open for write
staging: gasket: annotate ioctl arg with __user
staging: gasket: gasket_enable_dev fixups
staging: gasket: fix class create bug handling
staging: gasket: remove unnecessary code in coherent allocator
staging: gasket: gasket core error handling fixups
staging: gasket: don't release coherent mappings
staging: gasket: various cleanups

 MAINTAINERS                                |   1 +
 drivers/staging/gasket/Kconfig             |   2 +-
 drivers/staging/gasket/apex.h              |   7 +-
 drivers/staging/gasket/apex_driver.c       |  79 +++++++--------
 drivers/staging/gasket/gasket_core.c       | 107 ++++++++++-----------
 drivers/staging/gasket/gasket_core.h       |  11 ++-
 drivers/staging/gasket/gasket_ioctl.c      | 100 ++++++++++---------
 drivers/staging/gasket/gasket_ioctl.h      |   4 +-
 drivers/staging/gasket/gasket_page_table.c |  69 +++++++------
 drivers/staging/gasket/gasket_page_table.h |   8 +-
 drivers/staging/gasket/gasket_sysfs.c      |   4 +-
 11 files changed, 198 insertions(+), 194 deletions(-)
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#1 lore raw

PATCH 1/18 staging: gasket: remove X86 Kconfig restriction

From: Todd Poynor <[email protected]>

The gasket and apex drivers are to be used on other architectures besides X86.

---
 drivers/staging/gasket/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/gasket/Kconfig +1 -1
index c836389c1402..ef40c4c75e0f 100644
@@ -2,7 +2,7 @@ menu "Gasket devices"
 
 config STAGING_GASKET_FRAMEWORK
 	tristate "Gasket framework"
-	depends on PCI && X86_64
+	depends on PCI
 	help
 	  This framework supports Gasket-compatible devices, such as Apex.
 	  It is required for any of the following module(s).
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#2 lore raw

PATCH 2/18 MAINTAINERS: Add maintainer for drivers/staging/gasket

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Todd Poynor <[email protected]> will be a maintainer of the Gasket and Apex drivers.

---
 MAINTAINERS | 1 +
 1 file changed, 1 insertion(+)
MAINTAINERS +1 -0
index 562ee1e1cfd6..8f44e7bed49e 100644
@@ -5968,6 +5968,7 @@ GASKET DRIVER FRAMEWORK
 M:	Rob Springer <[email protected]>
 M:	John Joseph <[email protected]>
 M:	Ben Chan <[email protected]>
+M:	Todd Poynor <[email protected]>
 S:	Maintained
 F:	drivers/staging/gasket/
 
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#3 lore raw

Re: [PATCH 02/18] MAINTAINERS: Add maintainer for drivers/staging/gasket

Quoted 6 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Todd Poynor <[email protected]> will be a maintainer of the Gasket and Apex drivers.

I need an ack, or signed-off-by by the current maintainers that this is ok. And 4 maintainers is a lot, are you sure everyone here is the "main" contact?

thanks,

greg k-h

#4 lore raw

PATCH 3/18 staging: gasket: typo and whitespace cleanups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Trivial typo and whitespace fixes.

---
 drivers/staging/gasket/apex_driver.c       | 2 +-
 drivers/staging/gasket/gasket_core.h       | 2 +-
 drivers/staging/gasket/gasket_page_table.c | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)
drivers/staging/gasket/apex_driver.c +1 -1
index 395256704428..f2d082c06c98 100644
@@ -500,7 +500,7 @@ static int apex_enter_reset(struct gasket_dev *gasket_dev, uint type)
 					APEX_BAR2_REG_USER_HIB_DMA_PAUSED, 1, 1,
 					APEX_RESET_DELAY, APEX_RESET_RETRY)) {
 		gasket_log_error(gasket_dev,
-				 "DMAs did not quiece within timeout (%d ms)",
+				 "DMAs did not quiesce within timeout (%d ms)",
 				 APEX_RESET_RETRY * APEX_RESET_DELAY);
 		return -EINVAL;
 	}
drivers/staging/gasket/gasket_core.h +1 -1
index 5d6535a0f254..77d08c1265ca 100644
@@ -63,7 +63,7 @@ enum gasket_interrupt_type {
 
 /* Used to describe a Gasket interrupt. Contains an interrupt index, a register,
  * and packing data for that interrupt. The register and packing data
- * fields is relevant only for PCI_MSIX interrupt type and can be
+ * fields are relevant only for PCI_MSIX interrupt type and can be
  * set to 0 for everything else.
  */
 struct gasket_interrupt_desc {
drivers/staging/gasket/gasket_page_table.c +1 -1
index 5d3d33cac12f..d7600d8e385f 100644
@@ -354,7 +354,7 @@ int gasket_page_table_init(
 	    pg_tbl->config.mode == GASKET_PAGE_TABLE_MODE_SIMPLE) {
 		pg_tbl->num_simple_entries = total_entries;
 		pg_tbl->num_extended_entries = 0;
-		pg_tbl->extended_flag =	1ull << page_table_config->extended_bit;
+		pg_tbl->extended_flag = 1ull << page_table_config->extended_bit;
 	} else {
 		pg_tbl->num_simple_entries = 0;
 		pg_tbl->num_extended_entries = total_entries;
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#5 lore raw

Re: [PATCH 03/18] staging: gasket: typo and whitespace cleanups

Quoted 3 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Trivial typo and whitespace fixes.

typos are different from whitespace issues. Please break this up.

Yeah, it's a pain, sorry, but that's a requirement for kernel patches. It's to make it easier on the reviewer, not necessarily the creator of the patch :)

thanks,

greg k-h

#6 lore raw

PATCH 4/18 staging: gasket: device registration error and unregister fixups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Remove device registration on class creation fail.

Hold mutex around device removal updates for proper ordering of updates.

---
 drivers/staging/gasket/gasket_core.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/staging/gasket/gasket_core.c +7 -1
index 45914ebc8f44..3bdf7d36b397 100644
@@ -342,7 +342,8 @@ int gasket_register_device(const struct gasket_driver_desc *driver_desc)
 	if (IS_ERR_OR_NULL(internal->class)) {
 		gasket_nodev_error("Cannot register %s class [ret=%ld]",
 				   driver_desc->name, PTR_ERR(internal->class));
-		return PTR_ERR(internal->class);
+		ret = PTR_ERR(internal->class);
+		goto fail0;
 	}
 
 	/*
@@ -376,7 +377,10 @@ int gasket_register_device(const struct gasket_driver_desc *driver_desc)
 fail1:
 	class_destroy(internal->class);
 
+fail0:
+	mutex_lock(&g_mutex);
 	g_descs[desc_idx].driver_desc = NULL;
+	mutex_unlock(&g_mutex);
 	return ret;
 }
 EXPORT_SYMBOL(gasket_register_device);
@@ -413,7 +417,9 @@ void gasket_unregister_device(const struct gasket_driver_desc *driver_desc)
 	class_destroy(internal_desc->class);
 
 	/* Finally, effectively "remove" the driver. */
+	mutex_lock(&g_mutex);
 	g_descs[desc_idx].driver_desc = NULL;
+	mutex_unlock(&g_mutex);
 
 	gasket_nodev_info("removed %s driver", driver_desc->name);
 }
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#7 lore raw

Re: [PATCH 04/18] staging: gasket: device registration error and unregister fixups

drivers/staging/gasket/gasket_core.c int gasket_register_device(const struct gasket_driver_desc *driver_desc) · line 380

Quoted 5 lines · Todd Poynor
drivers/staging/gasket/gasket_core.c +1 -0
@@ -376,7 +377,10 @@ int gasket_register_device(const struct gasket_driver_desc *driver_desc)
 fail1:
 	class_destroy(internal->class);
 
+fail0:

GW-BASIC sytle label names are an anti-pattern. It's better to name the labels after what they do, just like function names describes what the function does. Here it would be "goto clear_desc;" or something. It doesn't have to be perfect but so long as you *try* to name it something useful that's better than not trying.

drivers/staging/gasket/gasket_core.c int gasket_register_device(const struct gasket_driver_desc *driver_desc) · line 385

Quoted 5 lines
drivers/staging/gasket/gasket_core.c +2 -0
+	mutex_lock(&g_mutex);
 	g_descs[desc_idx].driver_desc = NULL;
+	mutex_unlock(&g_mutex);
 	return ret;
 }

regards,
dan carpenter

#8 lore raw

Re: [PATCH 04/18] staging: gasket: device registration error and unregister fixups

Quoted 11 lines
Quoted 4 lines

fail1:

      class_destroy(internal->class);

+fail0:

GW-BASIC sytle label names are an anti-pattern. It's better to name the labels after what they do, just like function names describes what the function does. Here it would be "goto clear_desc;" or something. It doesn't have to be perfect but so long as you *try* to name it something useful that's better than not trying.

Quoted 8 lines

drivers/staging/gasket/gasket_core.c int gasket_register_device(const struct gasket_driver_desc *driver_desc) · line 385

Quoted 5 lines
drivers/staging/gasket/gasket_core.c +2 -0
+     mutex_lock(&g_mutex);
      g_descs[desc_idx].driver_desc = NULL;
+     mutex_unlock(&g_mutex);
      return ret;
 }

regards,
dan carpenter

Thanks, I'll fix up the newly-added one in a respin of this patch and fix up existing labels in a future patch.

Signature · 1 line
Todd
#9 lore raw

PATCH 5/18 staging: gasket: sysfs mapping creation fixups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Return EBUSY for attempt to create a mapping already in use.

Remove stale pointers on error allocating attr array.

---
 drivers/staging/gasket/gasket_sysfs.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/staging/gasket/gasket_sysfs.c +3 -1
index a3705d6e088a..d1856a09b894 100644
@@ -213,7 +213,7 @@ int gasket_sysfs_create_mapping(
 			"0x%p.", device);
 		put_mapping(mapping);
 		mutex_unlock(&function_mutex);
-		return -EINVAL;
+		return -EBUSY;
 	}
 
 	/* Find the first empty entry in the array. */
@@ -244,6 +244,8 @@ int gasket_sysfs_create_mapping(
 	mapping->attribute_count = 0;
 	if (!mapping->attributes) {
 		gasket_nodev_error("Unable to allocate sysfs attribute array.");
+		mapping->device = NULL;
+		mapping->gasket_dev = NULL;
 		mutex_unlock(&mapping->mutex);
 		mutex_unlock(&function_mutex);
 		return -ENOMEM;
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#10 lore raw

Re: [PATCH 05/18] staging: gasket: sysfs mapping creation fixups

Quoted 3 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Return EBUSY for attempt to create a mapping already in use.

Why?

Quoted 1 line

Remove stale pointers on error allocating attr array.

This should be a different patch.

thanks,

greg k-h

#11 lore raw

Re: [PATCH 05/18] staging: gasket: sysfs mapping creation fixups

Quoted 6 lines · Greg Kroah-Hartman
Quoted 3 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Return EBUSY for attempt to create a mapping already in use.

Why?

The existing code returns EINVAL which often means something bogus was requested, whereas EBUSY is sometimes used for requesting something already in use. If that's not an error typically returned in this situation I'm happy to drop it.

Quoted 4 lines
Quoted 2 lines

Remove stale pointers on error allocating attr array.

This should be a different patch.

Will do.

Quoted 4 lines

thanks,

greg k-h

Signature · 1 line
Todd
#12 lore raw

Re: [PATCH 05/18] staging: gasket: sysfs mapping creation fixups

Quoted 5 lines · Todd Poynor
Quoted 3 lines · Greg Kroah-Hartman
Quoted 2 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Return EBUSY for attempt to create a mapping already in use.

Why?

The existing code returns EINVAL which often means something bogus was requested, whereas EBUSY is sometimes used for requesting something already in use. If that's not an error typically returned in this situation I'm happy to drop it.

My "why" was rhetorical, please spell this out in the changelog text :)

thanks,

greg k-h

#13 lore raw

PATCH 6/18 staging: gasket: fix deadlock in pci driver unregister path

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

g_mutex held across pci_unregister_driver() call, also held in gasket_pci_remove(), which deadlocks.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
drivers/staging/gasket/gasket_core.c +2 -5
index 3bdf7d36b397..6d240dc59557 100644
@@ -668,13 +668,10 @@ static void gasket_pci_remove(struct pci_dev *pci_dev)
 	struct gasket_dev *gasket_dev = NULL;
 	const struct gasket_driver_desc *driver_desc;
 	/* Find the device desc. */
-	mutex_lock(&g_mutex);
+	__must_hold(&g_mutex);
 	internal_desc = lookup_internal_desc(pci_dev);
-	if (!internal_desc) {
-		mutex_unlock(&g_mutex);
+	if (!internal_desc)
 		return;
-	}
-	mutex_unlock(&g_mutex);
 
 	driver_desc = internal_desc->driver_desc;
 
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#14 lore raw

Re: [PATCH 06/18] staging: gasket: fix deadlock in pci driver unregister path

Quoted 4 lines · Todd Poynor

From: Todd Poynor <[email protected]>

g_mutex held across pci_unregister_driver() call, also held in gasket_pci_remove(), which deadlocks.

Which reminds me, why in the world do you all wrap pci register/unregister within your new subsystem? That's very odd, and should be reworked... Someone should go add that to the TODO list :)

thanks,

greg k-h

#15 lore raw

Re: [PATCH 06/18] staging: gasket: fix deadlock in pci driver unregister path

In reply to Todd Poynor
Quoted 18 lines · Todd Poynor

From: Todd Poynor <[email protected]>

g_mutex held across pci_unregister_driver() call, also held in gasket_pci_remove(), which deadlocks.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
drivers/staging/gasket/gasket_core.c +0 -0
index 3bdf7d36b397..6d240dc59557 100644

drivers/staging/gasket/gasket_core.c static void gasket_pci_remove(struct pci_dev *pci_dev) · line 671

Quoted 5 lines
drivers/staging/gasket/gasket_core.c +1 -1
@@ -668,13 +668,10 @@ static void gasket_pci_remove(struct pci_dev *pci_dev)
        struct gasket_dev *gasket_dev = NULL;
        const struct gasket_driver_desc *driver_desc;
        /* Find the device desc. */
-       mutex_lock(&g_mutex);
+       __must_hold(&g_mutex);

And what exactly ensures that mutex is held here? Yes, we are holding the mutex when we unload the driver, but PCI hot-unplug or unbinding the device though sysfs do not go through module unload code path, so you'll end up here without holding the mutex.

drivers/staging/gasket/gasket_core.c static void gasket_pci_remove(struct pci_dev *pci_dev) · line 676

Quoted 9 lines
drivers/staging/gasket/gasket_core.c +1 -4
        internal_desc = lookup_internal_desc(pci_dev);
-       if (!internal_desc) {
-               mutex_unlock(&g_mutex);
+       if (!internal_desc)
                return;
-       }
-       mutex_unlock(&g_mutex);
 
        driver_desc = internal_desc->driver_desc;

Thanks,
Dmitry

#16 lore raw

Re: [PATCH 06/18] staging: gasket: fix deadlock in pci driver unregister path

Quoted 14 lines · Dmitry Torokhov
Quoted 13 lines · Todd Poynor

From: Todd Poynor <[email protected]>

g_mutex held across pci_unregister_driver() call, also held in gasket_pci_remove(), which deadlocks.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
Quoted 15 lines

drivers/staging/gasket/gasket_core.c static void gasket_pci_remove(struct pci_dev *pci_dev) · line 671

Quoted 10 lines
drivers/staging/gasket/gasket_core.c +1 -1
index 3bdf7d36b397..6d240dc59557 100644
@@ -668,13 +668,10 @@ static void gasket_pci_remove(struct pci_dev *pci_dev)
        struct gasket_dev *gasket_dev = NULL;
        const struct gasket_driver_desc *driver_desc;
        /* Find the device desc. */
-       mutex_lock(&g_mutex);
+       __must_hold(&g_mutex);

And what exactly ensures that mutex is held here? Yes, we are holding the mutex when we unload the driver, but PCI hot-unplug or unbinding the device though sysfs do not go through module unload code path, so you'll end up here without holding the mutex.

Which is a huge reason the whole "wrap the pci core calls" is not going to work here at all. The device ownership rules are all wonky because of this. Unwinding that is key to getting all of this right.

thanks,

greg k-h

#17 lore raw

Re: [PATCH 06/18] staging: gasket: fix deadlock in pci driver unregister path

Quoted 31 lines · Greg Kroah-Hartman
Quoted 14 lines · Dmitry Torokhov
Quoted 13 lines · Todd Poynor

From: Todd Poynor <[email protected]>

g_mutex held across pci_unregister_driver() call, also held in gasket_pci_remove(), which deadlocks.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
Quoted 15 lines

drivers/staging/gasket/gasket_core.c static void gasket_pci_remove(struct pci_dev *pci_dev) · line 671

Quoted 10 lines
drivers/staging/gasket/gasket_core.c +1 -1
index 3bdf7d36b397..6d240dc59557 100644
@@ -668,13 +668,10 @@ static void gasket_pci_remove(struct pci_dev *pci_dev)
        struct gasket_dev *gasket_dev = NULL;
        const struct gasket_driver_desc *driver_desc;
        /* Find the device desc. */
-       mutex_lock(&g_mutex);
+       __must_hold(&g_mutex);

And what exactly ensures that mutex is held here? Yes, we are holding the mutex when we unload the driver, but PCI hot-unplug or unbinding the device though sysfs do not go through module unload code path, so you'll end up here without holding the mutex.

Quoted 3 lines

Which is a huge reason the whole "wrap the pci core calls" is not going to work here at all. The device ownership rules are all wonky because of this. Unwinding that is key to getting all of this right.

OK, I'll drop this patch in favor of redoing things not to wrap PCI core calls in the future, thanks.

Quoted 4 lines

thanks,

greg k-h

Signature · 1 line
Todd
#18 lore raw

PATCH 7/18 staging: gasket: convert gasket_mmap_has_permissions to bool return

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

gasket_mmap_has_permissions() should return a boolean value.

---
 drivers/staging/gasket/gasket_core.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/staging/gasket/gasket_core.c +6 -6
index 6d240dc59557..155c69446f0b 100644
@@ -1245,19 +1245,19 @@ static int gasket_release(struct inode *inode, struct file *file)
  * that the provided descriptor/range is of adequate size to hold the range to
  * be mapped.
  */
-static int gasket_mmap_has_permissions(
+static bool gasket_mmap_has_permissions(
 	struct gasket_dev *gasket_dev, struct vm_area_struct *vma,
 	int bar_permissions)
 {
 	int requested_permissions;
 	/* Always allow sysadmin to access. */
 	if (capable(CAP_SYS_ADMIN))
-		return 1;
+		return true;
 
 	/* Never allow non-sysadmins to access to a dead device. */
 	if (gasket_dev->status != GASKET_STATUS_ALIVE) {
 		gasket_log_info(gasket_dev, "Device is dead.");
-		return 0;
+		return false;
 	}
 
 	/* Make sure that no wrong flags are set. */
@@ -1269,7 +1269,7 @@ static int gasket_mmap_has_permissions(
 			"Attempting to map a region with requested permissions "
 			"0x%x, but region has permissions 0x%x.",
 			requested_permissions, bar_permissions);
-		return 0;
+		return false;
 	}
 
 	/* Do not allow a non-owner to write. */
@@ -1279,10 +1279,10 @@ static int gasket_mmap_has_permissions(
 			gasket_dev,
 			"Attempting to mmap a region for write without owning "
 			"device.");
-		return 0;
+		return false;
 	}
 
-	return 1;
+	return true;
 }
 
 /*
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#19 lore raw

PATCH 8/18 staging: gasket: gasket_wait_with_reschedule fixups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Return -ETIMEDOUT on timeouts.

Don't need 64-bit sized retry count.

Use msleep().

Return immediately when condition hit.

---
 drivers/staging/gasket/apex_driver.c |  8 ++++----
 drivers/staging/gasket/gasket_core.c | 11 ++++++-----
 drivers/staging/gasket/gasket_core.h |  2 +-
 3 files changed, 11 insertions(+), 10 deletions(-)
drivers/staging/gasket/apex_driver.c +4 -4
index f2d082c06c98..b1318482ba65 100644
@@ -502,7 +502,7 @@ static int apex_enter_reset(struct gasket_dev *gasket_dev, uint type)
 		gasket_log_error(gasket_dev,
 				 "DMAs did not quiesce within timeout (%d ms)",
 				 APEX_RESET_RETRY * APEX_RESET_DELAY);
-		return -EINVAL;
+		return -ETIMEDOUT;
 	}
 
 	/*  - Enable GCB reset (0x1 to rg_rst_gcb) */
@@ -525,7 +525,7 @@ static int apex_enter_reset(struct gasket_dev *gasket_dev, uint type)
 			gasket_dev,
 			"RAM did not shut down within timeout (%d ms)",
 			APEX_RESET_RETRY * APEX_RESET_DELAY);
-		return -EINVAL;
+		return -ETIMEDOUT;
 	}
 
 	return 0;
@@ -576,7 +576,7 @@ static int apex_quit_reset(struct gasket_dev *gasket_dev, uint type)
 			gasket_dev,
 			"RAM did not enable within timeout (%d ms)",
 			APEX_RESET_RETRY * APEX_RESET_DELAY);
-		return -EINVAL;
+		return -ETIMEDOUT;
 	}
 
 	/*    - Wait for Reset complete. */
@@ -588,7 +588,7 @@ static int apex_quit_reset(struct gasket_dev *gasket_dev, uint type)
 			gasket_dev,
 			"GCB did not leave reset within timeout (%d ms)",
 			APEX_RESET_RETRY * APEX_RESET_DELAY);
-		return -EINVAL;
+		return -ETIMEDOUT;
 	}
 
 	if (!allow_hw_clock_gating) {
drivers/staging/gasket/gasket_core.c +6 -5
index 155c69446f0b..6316e6c800ba 100644
@@ -21,6 +21,7 @@
 #include "gasket_page_table.h"
 #include "gasket_sysfs.h"
 
+#include <linux/delay.h>
 #include <linux/fs.h>
 #include <linux/init.h>
 #include <linux/of.h>
@@ -2135,16 +2136,16 @@ EXPORT_SYMBOL(gasket_wait_sync);
  **/
 int gasket_wait_with_reschedule(
 	struct gasket_dev *gasket_dev, int bar, u64 offset, u64 mask, u64 val,
-	u64 max_retries, u64 delay_ms)
+	uint max_retries, u64 delay_ms)
 {
-	u64 retries = 0;
+	uint retries = 0;
 	u64 tmp;
 
 	while (retries < max_retries) {
 		tmp = gasket_dev_read_64(gasket_dev, bar, offset);
 		if ((tmp & mask) == val)
-			break;
-		schedule_timeout(msecs_to_jiffies(delay_ms));
+			return 0;
+		msleep(delay_ms);
 		retries++;
 	}
 	if (retries == max_retries) {
@@ -2152,7 +2153,7 @@ int gasket_wait_with_reschedule(
 			gasket_dev,
 			"gasket_wait_with_reschedule timeout: reg %llx timeout (%llu ms)",
 			offset, max_retries * delay_ms);
-		return -EINVAL;
+		return -ETIMEDOUT;
 	}
 	return 0;
 }
drivers/staging/gasket/gasket_core.h +1 -1
index 77d08c1265ca..16bee478dce6 100644
@@ -714,6 +714,6 @@ int gasket_wait_sync(
 /* Helper function, Asynchronous waits on a given set of bits. */
 int gasket_wait_with_reschedule(
 	struct gasket_dev *gasket_dev, int bar, u64 offset, u64 mask, u64 val,
-	u64 max_retries, u64 delay_ms);
+	uint max_retries, u64 delay_ms);
 
 #endif /* __GASKET_CORE_H__ */
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#20 lore raw

Re: [PATCH 08/18] staging: gasket: gasket_wait_with_reschedule fixups

Quoted 9 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Return -ETIMEDOUT on timeouts.

Don't need 64-bit sized retry count.

Use msleep().

Return immediately when condition hit.

Huge hint, when you have to list the different things you do in a patch, that means it should be split up into smaller patches.

Please rework this patch series based on this. I'll stop reviewing them now and wait for a new set of patches.

thanks,

greg k-h

#21 lore raw

PATCH 9/18 staging: gasket: bail out of reset sequence on device callback error

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

If device reset callback returns an error, error out at the gasket level.

---
 drivers/staging/gasket/gasket_core.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/staging/gasket/gasket_core.c +3 -1
index 6316e6c800ba..ffd6ce801313 100644
@@ -1879,9 +1879,11 @@ int gasket_reset_nolock(struct gasket_dev *gasket_dev, uint reset_type)
 
 	/* Perform a device reset of the requested type. */
 	ret = driver_desc->device_reset_cb(gasket_dev, reset_type);
-	if (ret)
+	if (ret) {
 		gasket_log_error(
 			gasket_dev, "Device reset cb returned %d.", ret);
+		return ret;
+	}
 
 	/* Reinitialize the page tables and interrupt framework. */
 	for (i = 0; i < driver_desc->num_page_tables; ++i)
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#22 lore raw

PATCH 10/18 staging: gasket: gasket_open use container_of()

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Use container_of(), drop unnecessary NULL check.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
drivers/staging/gasket/gasket_core.c +2 -5
index ffd6ce801313..0c45c54254fb 100644
@@ -1099,12 +1099,9 @@ static int gasket_open(struct inode *inode, struct file *filp)
 	const struct gasket_driver_desc *driver_desc;
 	struct gasket_ownership *ownership;
 	char task_name[TASK_COMM_LEN];
-	struct gasket_cdev_info *dev_info = gasket_cdev_get_info(inode->i_cdev);
+	struct gasket_cdev_info *dev_info =
+	    container_of(inode->i_cdev, struct gasket_cdev_info, cdev);
 
-	if (!dev_info) {
-		gasket_nodev_error("Unable to retrieve device data");
-		return -EINVAL;
-	}
 	gasket_dev = dev_info->gasket_dev_ptr;
 	driver_desc = gasket_dev->internal_desc->driver_desc;
 	ownership = &dev_info->ownership;
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#23 lore raw

Re: [PATCH 10/18] staging: gasket: gasket_open use container_of()

Quoted 17 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Use container_of(), drop unnecessary NULL check.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
drivers/staging/gasket/gasket_core.c +0 -0
index ffd6ce801313..0c45c54254fb 100644

drivers/staging/gasket/gasket_core.c static int gasket_open(struct inode *inode, struct file *filp) · line 1102

Quoted 4 lines
drivers/staging/gasket/gasket_core.c +0 -1
@@ -1099,12 +1099,9 @@ static int gasket_open(struct inode *inode, struct file *filp)
        const struct gasket_driver_desc *driver_desc;
        struct gasket_ownership *ownership;
        char task_name[TASK_COMM_LEN];
-       struct gasket_cdev_info *dev_info = gasket_cdev_get_info(inode->i_cdev);

Are there other users of gasket_cdev_get_info()? if this was the only one we should remove the macro,otherwise we should convert them as well.

Quoted 13 lines
drivers/staging/gasket/gasket_core.c +2 -5
+       struct gasket_cdev_info *dev_info =
+           container_of(inode->i_cdev, struct gasket_cdev_info, cdev);
 
-       if (!dev_info) {
-               gasket_nodev_error("Unable to retrieve device data");
-               return -EINVAL;
-       }
        gasket_dev = dev_info->gasket_dev_ptr;
        driver_desc = gasket_dev->internal_desc->driver_desc;
        ownership = &dev_info->ownership;
--

2.18.0.203.gfac676dfb9-goog

#24 lore raw

Re: [PATCH 10/18] staging: gasket: gasket_open use container_of()

Quoted 13 lines · Dmitry Torokhov
Quoted 12 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Use container_of(), drop unnecessary NULL check.

---
 drivers/staging/gasket/gasket_core.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
Quoted 13 lines

drivers/staging/gasket/gasket_core.c static int gasket_open(struct inode *inode, struct file *filp) · line 1102

Quoted 9 lines
drivers/staging/gasket/gasket_core.c +0 -1
index ffd6ce801313..0c45c54254fb 100644
@@ -1099,12 +1099,9 @@ static int gasket_open(struct inode *inode, struct file *filp)
        const struct gasket_driver_desc *driver_desc;
        struct gasket_ownership *ownership;
        char task_name[TASK_COMM_LEN];
-       struct gasket_cdev_info *dev_info = gasket_cdev_get_info(inode->i_cdev);

Are there other users of gasket_cdev_get_info()? if this was the only one we should remove the macro,otherwise we should convert them as well.

No other users, I'll add a patch to remove it, thanks.

Quoted 14 lines
Quoted 14 lines
drivers/staging/gasket/gasket_core.c +2 -5
+       struct gasket_cdev_info *dev_info =
+           container_of(inode->i_cdev, struct gasket_cdev_info, cdev);
 
-       if (!dev_info) {
-               gasket_nodev_error("Unable to retrieve device data");
-               return -EINVAL;
-       }
        gasket_dev = dev_info->gasket_dev_ptr;
        driver_desc = gasket_dev->internal_desc->driver_desc;
        ownership = &dev_info->ownership;
--

2.18.0.203.gfac676dfb9-goog

Signature · 1 line
Todd
#25 lore raw

PATCH 11/18 staging: gasket: always allow root open for write

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

---
 drivers/staging/gasket/apex_driver.c  |  9 +++------
 drivers/staging/gasket/gasket_core.c  |  8 +++++---
 drivers/staging/gasket/gasket_ioctl.c | 15 ++++++---------
 3 files changed, 14 insertions(+), 18 deletions(-)
drivers/staging/gasket/apex_driver.c +3 -6
index b1318482ba65..ffe11d8168ea 100644
@@ -644,13 +644,10 @@ static bool is_gcb_in_reset(struct gasket_dev *gasket_dev)
 static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 {
 	struct gasket_dev *gasket_dev = filp->private_data;
-	int root = capable(CAP_SYS_ADMIN);
-	int is_owner = gasket_dev->dev_info.ownership.is_owned &&
-		       current->tgid == gasket_dev->dev_info.ownership.owner;
+	fmode_t write;
 
-	if (root || is_owner)
-		return 1;
-	return 0;
+	write = filp->f_mode & FMODE_WRITE;
+	return write;
 }
 
 /*
drivers/staging/gasket/gasket_core.c +5 -3
index 0c45c54254fb..5908964f0039 100644
@@ -1101,6 +1101,7 @@ static int gasket_open(struct inode *inode, struct file *filp)
 	char task_name[TASK_COMM_LEN];
 	struct gasket_cdev_info *dev_info =
 	    container_of(inode->i_cdev, struct gasket_cdev_info, cdev);
+	int is_root = capable(CAP_SYS_ADMIN);
 
 	gasket_dev = dev_info->gasket_dev_ptr;
 	driver_desc = gasket_dev->internal_desc->driver_desc;
@@ -1114,7 +1115,7 @@ static int gasket_open(struct inode *inode, struct file *filp)
 		"Attempting to open with tgid %u (%s) (f_mode: 0%03o, "
 		"fmode_write: %d is_root: %u)",
 		current->tgid, task_name, filp->f_mode,
-		(filp->f_mode & FMODE_WRITE), capable(CAP_SYS_ADMIN));
+		(filp->f_mode & FMODE_WRITE), is_root);
 
 	/* Always allow non-writing accesses. */
 	if (!(filp->f_mode & FMODE_WRITE)) {
@@ -1128,8 +1129,9 @@ static int gasket_open(struct inode *inode, struct file *filp)
 		gasket_dev, "Current owner open count (owning tgid %u): %d.",
 		ownership->owner, ownership->write_open_count);
 
-	/* Opening a node owned by another TGID is an error (even root.) */
-	if (ownership->is_owned && ownership->owner != current->tgid) {
+	/* Opening a node owned by another TGID is an error (unless root) */
+	if (ownership->is_owned && ownership->owner != current->tgid &&
+	    !is_root) {
 		gasket_log_error(
 			gasket_dev,
 			"Process %u is opening a node held by %u.",
drivers/staging/gasket/gasket_ioctl.c +6 -9
index 4758083fb19b..23875fd465f6 100644
@@ -180,7 +180,7 @@ long gasket_is_supported_ioctl(uint cmd)
  */
 static uint gasket_ioctl_check_permissions(struct file *filp, uint cmd)
 {
-	uint alive, root, device_owner;
+	uint alive;
 	fmode_t read, write;
 	struct gasket_dev *gasket_dev = (struct gasket_dev *)filp->private_data;
 
@@ -191,33 +191,30 @@ static uint gasket_ioctl_check_permissions(struct file *filp, uint cmd)
 			alive, gasket_dev->status);
 	}
 
-	root = capable(CAP_SYS_ADMIN);
 	read = filp->f_mode & FMODE_READ;
 	write = filp->f_mode & FMODE_WRITE;
-	device_owner = (gasket_dev->dev_info.ownership.is_owned &&
-			current->tgid == gasket_dev->dev_info.ownership.owner);
 
 	switch (cmd) {
 	case GASKET_IOCTL_RESET:
 	case GASKET_IOCTL_CLEAR_INTERRUPT_COUNTS:
-		return root || (write && device_owner);
+		return write;
 
 	case GASKET_IOCTL_PAGE_TABLE_SIZE:
 	case GASKET_IOCTL_SIMPLE_PAGE_TABLE_SIZE:
 	case GASKET_IOCTL_NUMBER_PAGE_TABLES:
-		return root || read;
+		return read;
 
 	case GASKET_IOCTL_PARTITION_PAGE_TABLE:
 	case GASKET_IOCTL_CONFIG_COHERENT_ALLOCATOR:
-		return alive && (root || (write && device_owner));
+		return alive && write;
 
 	case GASKET_IOCTL_MAP_BUFFER:
 	case GASKET_IOCTL_UNMAP_BUFFER:
-		return alive && (root || (write && device_owner));
+		return alive && write;
 
 	case GASKET_IOCTL_CLEAR_EVENTFD:
 	case GASKET_IOCTL_SET_EVENTFD:
-		return alive && (root || (write && device_owner));
+		return alive && write;
 	}
 
 	return 0; /* unknown permissions */
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
Reviewed-by on v1 11/18
#26 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 10 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

I think this patch is good as is, but as a followup you should create a patch that supports user namespaces, i.e. replaces capable(CAP_SYS_ADMIN) with ns_capable(...) in gasket_open() so you can allow containers to control the device, if necessary.

Thanks,
Dmitry

#27 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 8 lines · Dmitry Torokhov
Quoted 7 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

Quoted 8 lines
Quoted 3 lines

I think this patch is good as is, but as a followup you should create a patch that supports user namespaces, i.e. replaces capable(CAP_SYS_ADMIN) with ns_capable(...) in gasket_open() so you can allow containers to control the device, if necessary.

Thanks, I'll add that to the list.

Quoted 5 lines

Thanks,
Dmitry

Signature · 1 line
Todd
#28 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

In reply to Todd Poynor
Quoted 22 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

---
 drivers/staging/gasket/apex_driver.c  |  9 +++------
 drivers/staging/gasket/gasket_core.c  |  8 +++++---
 drivers/staging/gasket/gasket_ioctl.c | 15 ++++++---------
 3 files changed, 14 insertions(+), 18 deletions(-)
drivers/staging/gasket/apex_driver.c +0 -0
index b1318482ba65..ffe11d8168ea 100644
@@ -644,13 +644,10 @@ static bool is_gcb_in_reset(struct gasket_dev *gasket_dev)
 static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 {

drivers/staging/gasket/apex_driver.c static bool is_gcb_in_reset(struct gasket_dev *gasket_dev) · line 649

Quoted 10 lines
drivers/staging/gasket/apex_driver.c +2 -6
 	struct gasket_dev *gasket_dev = filp->private_data;
-	int root = capable(CAP_SYS_ADMIN);
-	int is_owner = gasket_dev->dev_info.ownership.is_owned &&
-		       current->tgid == gasket_dev->dev_info.ownership.owner;
+	fmode_t write;
 
-	if (root || is_owner)
-		return 1;
-	return 0;
+	write = filp->f_mode & FMODE_WRITE;

Ok, this is insane. You don't change, or check, the permissions on a file handle while it is already open, as you only check the permissions on OPEN, not on WRITE. See the recent rant from Linus on the linux-api list for yet-another-long-threaad in which he explains this.

So this whole ioctl can just be removed, it is totally crazy and wrong and should just be removed.

Want me to go remove it right now?

thanks,

greg k-h

#29 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 40 lines · Greg Kroah-Hartman
Quoted 22 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

---
 drivers/staging/gasket/apex_driver.c  |  9 +++------
 drivers/staging/gasket/gasket_core.c  |  8 +++++---
 drivers/staging/gasket/gasket_ioctl.c | 15 ++++++---------
 3 files changed, 14 insertions(+), 18 deletions(-)
drivers/staging/gasket/apex_driver.c +0 -0
index b1318482ba65..ffe11d8168ea 100644
@@ -644,13 +644,10 @@ static bool is_gcb_in_reset(struct gasket_dev *gasket_dev)
 static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 {

drivers/staging/gasket/apex_driver.c static bool is_gcb_in_reset(struct gasket_dev *gasket_dev) · line 649

Quoted 10 lines
drivers/staging/gasket/apex_driver.c +2 -6
      struct gasket_dev *gasket_dev = filp->private_data;
-     int root = capable(CAP_SYS_ADMIN);
-     int is_owner = gasket_dev->dev_info.ownership.is_owned &&
-                    current->tgid == gasket_dev->dev_info.ownership.owner;
+     fmode_t write;
 
-     if (root || is_owner)
-             return 1;
-     return 0;
+     write = filp->f_mode & FMODE_WRITE;

Ok, this is insane. You don't change, or check, the permissions on a file handle while it is already open, as you only check the permissions on OPEN, not on WRITE. See the recent rant from Linus on the linux-api list for yet-another-long-threaad in which he explains this.

Quoted 2 lines

So this whole ioctl can just be removed, it is totally crazy and wrong and should just be removed.

No, the code checks whether the requested ioctl command is compatible with the mode the file handle was open with. There are some ioctls that are allowed on file handle opened for read and others that require file handle to be opened for write. That is all. It does not change permissions past open.

Thanks,
Dmitry

#30 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 42 lines · Dmitry Torokhov
Quoted 40 lines · Greg Kroah-Hartman
Quoted 22 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

---
 drivers/staging/gasket/apex_driver.c  |  9 +++------
 drivers/staging/gasket/gasket_core.c  |  8 +++++---
 drivers/staging/gasket/gasket_ioctl.c | 15 ++++++---------
 3 files changed, 14 insertions(+), 18 deletions(-)
drivers/staging/gasket/apex_driver.c +0 -0
index b1318482ba65..ffe11d8168ea 100644
@@ -644,13 +644,10 @@ static bool is_gcb_in_reset(struct gasket_dev *gasket_dev)
 static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 {

drivers/staging/gasket/apex_driver.c static bool is_gcb_in_reset(struct gasket_dev *gasket_dev) · line 649

Quoted 10 lines
drivers/staging/gasket/apex_driver.c +2 -6
      struct gasket_dev *gasket_dev = filp->private_data;
-     int root = capable(CAP_SYS_ADMIN);
-     int is_owner = gasket_dev->dev_info.ownership.is_owned &&
-                    current->tgid == gasket_dev->dev_info.ownership.owner;
+     fmode_t write;
 
-     if (root || is_owner)
-             return 1;
-     return 0;
+     write = filp->f_mode & FMODE_WRITE;

Ok, this is insane. You don't change, or check, the permissions on a file handle while it is already open, as you only check the permissions on OPEN, not on WRITE. See the recent rant from Linus on the linux-api list for yet-another-long-threaad in which he explains this.

Quoted 8 lines
Quoted 2 lines

So this whole ioctl can just be removed, it is totally crazy and wrong and should just be removed.

No, the code checks whether the requested ioctl command is compatible with the mode the file handle was open with. There are some ioctls that are allowed on file handle opened for read and others that require file handle to be opened for write. That is all. It does not change permissions past open.

That's really not obvious here :)

And is odd on a whole other set of "crazy design", but ok, let's let it live for now.

I can't wait for people to just realize this whole "new" subsystem can be replaced with UIO, but that's a topic for a different thread...

thanks,

greg k-h

#31 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 53 lines · Greg Kroah-Hartman
Quoted 42 lines · Dmitry Torokhov
Quoted 40 lines · Greg Kroah-Hartman
Quoted 22 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Always allow root to open device for writing.

Drop special-casing of ioctl permissions for root vs. owner.

---
 drivers/staging/gasket/apex_driver.c  |  9 +++------
 drivers/staging/gasket/gasket_core.c  |  8 +++++---
 drivers/staging/gasket/gasket_ioctl.c | 15 ++++++---------
 3 files changed, 14 insertions(+), 18 deletions(-)
drivers/staging/gasket/apex_driver.c +0 -0
index b1318482ba65..ffe11d8168ea 100644
@@ -644,13 +644,10 @@ static bool is_gcb_in_reset(struct gasket_dev *gasket_dev)
 static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 {

drivers/staging/gasket/apex_driver.c static bool is_gcb_in_reset(struct gasket_dev *gasket_dev) · line 649

Quoted 10 lines
drivers/staging/gasket/apex_driver.c +2 -6
      struct gasket_dev *gasket_dev = filp->private_data;
-     int root = capable(CAP_SYS_ADMIN);
-     int is_owner = gasket_dev->dev_info.ownership.is_owned &&
-                    current->tgid == gasket_dev->dev_info.ownership.owner;
+     fmode_t write;
 
-     if (root || is_owner)
-             return 1;
-     return 0;
+     write = filp->f_mode & FMODE_WRITE;

Ok, this is insane. You don't change, or check, the permissions on a file handle while it is already open, as you only check the permissions on OPEN, not on WRITE. See the recent rant from Linus on the linux-api list for yet-another-long-threaad in which he explains this.

Quoted 8 lines
Quoted 2 lines

So this whole ioctl can just be removed, it is totally crazy and wrong and should just be removed.

No, the code checks whether the requested ioctl command is compatible with the mode the file handle was open with. There are some ioctls that are allowed on file handle opened for read and others that require file handle to be opened for write. That is all. It does not change permissions past open.

Quoted 4 lines

That's really not obvious here :)

And is odd on a whole other set of "crazy design", but ok, let's let it live for now.

Are you talking about ioctl still or something else? Because such ioctl handling is quite common, take a look at blkdev for example, where disacrd ioctl is only allowed when device is open for writing.

Quoted 3 lines

I can't wait for people to just realize this whole "new" subsystem can be replaced with UIO, but that's a topic for a different thread...

Yes, that is true and that is why I am not sure why we are going through all this staging exercise.

As far as I understand we'd still need to have quite a bit of kernel code so that we can safely program DMA controller (it does not look like uio_dmem_genirq.c is sufficient as is for gasket needs), but that should be solvable.

Thanks,
Dmitry

#32 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 10 lines · Dmitry Torokhov
Quoted 2 lines

I can't wait for people to just realize this whole "new" subsystem can be replaced with UIO, but that's a topic for a different thread...

Yes, that is true and that is why I am not sure why we are going through all this staging exercise.

As far as I understand we'd still need to have quite a bit of kernel code so that we can safely program DMA controller (it does not look like uio_dmem_genirq.c is sufficient as is for gasket needs), but that should be solvable.

I agree, it should be solvable, and much smaller and simpler than this whole large chunk of "subsystem+driver" code. But I'm not the one having to do this work, and it provides a bunch of easy cleanups for people looking to get into kernel development, so I don't mind :)

But the "maintainers" should keep this in mind, as it is, this code is _not_ acceptable for the main kernel tree because of the UIO framework already present.

thanks,

greg k-h

#33 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 12 lines · Greg Kroah-Hartman
Quoted 10 lines · Dmitry Torokhov
Quoted 2 lines

I can't wait for people to just realize this whole "new" subsystem can be replaced with UIO, but that's a topic for a different thread...

Yes, that is true and that is why I am not sure why we are going through all this staging exercise.

As far as I understand we'd still need to have quite a bit of kernel code so that we can safely program DMA controller (it does not look like uio_dmem_genirq.c is sufficient as is for gasket needs), but that should be solvable.

Quoted 8 lines

I agree, it should be solvable, and much smaller and simpler than this whole large chunk of "subsystem+driver" code. But I'm not the one having to do this work, and it provides a bunch of easy cleanups for people looking to get into kernel development, so I don't mind :)

But the "maintainers" should keep this in mind, as it is, this code is _not_ acceptable for the main kernel tree because of the UIO framework already present.

My own preference is to rewrite the apex driver entirely in-kernel and pull in its userspace parts here. If I don't receive significant pushback on that I'll start doing that real soon.

Quoted 4 lines

thanks,

greg k-h

Signature · 1 line
Todd
#34 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 19 lines · Todd Poynor
Quoted 17 lines · Greg Kroah-Hartman
Quoted 10 lines · Dmitry Torokhov
Quoted 2 lines

I can't wait for people to just realize this whole "new" subsystem can be replaced with UIO, but that's a topic for a different thread...

Yes, that is true and that is why I am not sure why we are going through all this staging exercise.

As far as I understand we'd still need to have quite a bit of kernel code so that we can safely program DMA controller (it does not look like uio_dmem_genirq.c is sufficient as is for gasket needs), but that should be solvable.

I agree, it should be solvable, and much smaller and simpler than this whole large chunk of "subsystem+driver" code. But I'm not the one having to do this work, and it provides a bunch of easy cleanups for people looking to get into kernel development, so I don't mind :)

Quoted 7 lines
Quoted 3 lines

But the "maintainers" should keep this in mind, as it is, this code is _not_ acceptable for the main kernel tree because of the UIO framework already present.

My own preference is to rewrite the apex driver entirely in-kernel and pull in its userspace parts here. If I don't receive significant pushback on that I'll start doing that real soon.

Why would we object to that?

#35 lore raw

Re: [PATCH 11/18] staging: gasket: always allow root open for write

Quoted 8 lines · Greg Kroah-Hartman
Quoted 1 line · Todd Poynor

On Sun, Jul 15, 2018 at 3:03 AM, Greg Kroah-Hartman

...

Quoted 3 lines

My own preference is to rewrite the apex driver entirely in-kernel and pull in its userspace parts here. If I don't receive significant pushback on that I'll start doing that real soon.

Why would we object to that?

I've seen folks that have a preference for UIO whenever possible, but I'm of the "just do kernel-like stuff in the kernel" camp. There are folks cc'ed who may have strong opinions about the subject for various reasons; absent any serious objections I'll put that on a longer-term todo list. Thanks,

Signature · 1 line
Todd
#36 lore raw

PATCH 12/18 staging: gasket: annotate ioctl arg with __user

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

For sparse checking.

---
 drivers/staging/gasket/apex_driver.c  | 12 ++---
 drivers/staging/gasket/gasket_core.c  |  6 ++-
 drivers/staging/gasket/gasket_core.h  |  4 +-
 drivers/staging/gasket/gasket_ioctl.c | 72 ++++++++++++++-------------
 drivers/staging/gasket/gasket_ioctl.h |  4 +-
 5 files changed, 52 insertions(+), 46 deletions(-)
drivers/staging/gasket/apex_driver.c +6 -6
index ffe11d8168ea..40af91952283 100644
@@ -12,6 +12,7 @@
  * GNU General Public License for more details.
  */
 
+#include <linux/compiler.h>
 #include <linux/delay.h>
 #include <linux/fs.h>
 #include <linux/init.h>
@@ -149,9 +150,9 @@ static int apex_get_status(struct gasket_dev *gasket_dev);
 
 static uint apex_ioctl_check_permissions(struct file *file, uint cmd);
 
-static long apex_ioctl(struct file *file, uint cmd, ulong arg);
+static long apex_ioctl(struct file *file, uint cmd, void __user *arg);
 
-static long apex_clock_gating(struct gasket_dev *gasket_dev, ulong arg);
+static long apex_clock_gating(struct gasket_dev *gasket_dev, void __user *arg);
 
 static int apex_enter_reset(struct gasket_dev *gasket_dev, uint type);
 
@@ -643,7 +644,6 @@ static bool is_gcb_in_reset(struct gasket_dev *gasket_dev)
  */
 static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 {
-	struct gasket_dev *gasket_dev = filp->private_data;
 	fmode_t write;
 
 	write = filp->f_mode & FMODE_WRITE;
@@ -653,7 +653,7 @@ static uint apex_ioctl_check_permissions(struct file *filp, uint cmd)
 /*
  * Apex-specific ioctl handler.
  */
-static long apex_ioctl(struct file *filp, uint cmd, ulong arg)
+static long apex_ioctl(struct file *filp, uint cmd, void __user *arg)
 {
 	struct gasket_dev *gasket_dev = filp->private_data;
 
@@ -673,7 +673,7 @@ static long apex_ioctl(struct file *filp, uint cmd, ulong arg)
  * @gasket_dev: Gasket device pointer.
  * @arg: User ioctl arg, in this case to a apex_gate_clock_ioctl struct.
  */
-static long apex_clock_gating(struct gasket_dev *gasket_dev, ulong arg)
+static long apex_clock_gating(struct gasket_dev *gasket_dev, void __user *arg)
 {
 	struct apex_gate_clock_ioctl ibuf;
 
@@ -681,7 +681,7 @@ static long apex_clock_gating(struct gasket_dev *gasket_dev, ulong arg)
 		return 0;
 
 	if (allow_sw_clock_gating) {
-		if (copy_from_user(&ibuf, (void __user *)arg, sizeof(ibuf)))
+		if (copy_from_user(&ibuf, arg, sizeof(ibuf)))
 			return -EFAULT;
 
 		gasket_log_error(
drivers/staging/gasket/gasket_core.c +4 -2
index 5908964f0039..65aa7cf454fb 100644
@@ -21,6 +21,7 @@
 #include "gasket_page_table.h"
 #include "gasket_sysfs.h"
 
+#include <linux/compiler.h>
 #include <linux/delay.h>
 #include <linux/fs.h>
 #include <linux/init.h>
@@ -1842,14 +1843,15 @@ static long gasket_ioctl(struct file *filp, uint cmd, ulong arg)
 		 * check_and_invoke_callback.
 		 */
 		if (driver_desc->ioctl_handler_cb)
-			return driver_desc->ioctl_handler_cb(filp, cmd, arg);
+			return driver_desc->ioctl_handler_cb(
+			    filp, cmd, (void __user *)arg);
 
 		gasket_log_error(
 			gasket_dev, "Received unknown ioctl 0x%x", cmd);
 		return -EINVAL;
 	}
 
-	return gasket_handle_ioctl(filp, cmd, arg);
+	return gasket_handle_ioctl(filp, cmd, (void __user *)arg);
 }
 
 int gasket_reset(struct gasket_dev *gasket_dev, uint reset_type)
drivers/staging/gasket/gasket_core.h +2 -2
index 16bee478dce6..177b239e2c50 100644
@@ -322,7 +322,7 @@ struct gasket_dev {
 
 /* Type of the ioctl permissions check callback. See below. */
 typedef int (*gasket_ioctl_permissions_cb_t)(
-	struct file *filp, uint cmd, ulong arg);
+	struct file *filp, uint cmd, void __user *arg);
 
 /*
  * Device type descriptor.
@@ -556,7 +556,7 @@ struct gasket_driver_desc {
 	 * return -EINVAL. Should return an error status (either -EINVAL or
 	 * the error result of the ioctl being handled).
 	 */
-	long (*ioctl_handler_cb)(struct file *filp, uint cmd, ulong arg);
+	long (*ioctl_handler_cb)(struct file *filp, uint cmd, void __user *arg);
 
 	/*
 	 * device_status_cb: Callback to determine device health.
drivers/staging/gasket/gasket_ioctl.c +37 -35

200 lines not on this page · raw message

drivers/staging/gasket/gasket_ioctl.h +3 -1
index df868000c803..a9e10e4bc84a 100644
@@ -14,6 +14,8 @@
 
 #include "gasket_core.h"
 
+#include <linux/compiler.h>
+
 /*
  * Handle Gasket common ioctls.
  * @filp: Pointer to the ioctl's file.
@@ -22,7 +24,7 @@
  *
  * Returns 0 on success and nonzero on failure.
  */
-long gasket_handle_ioctl(struct file *filp, uint cmd, ulong arg);
+long gasket_handle_ioctl(struct file *filp, uint cmd, void __user *arg);
 
 /*
  * Determines if an ioctl is part of the standard Gasket framework.
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#37 lore raw

Re: [PATCH 12/18] staging: gasket: annotate ioctl arg with __user

Quoted 8 lines · Todd Poynor

From: Todd Poynor <[email protected]>

For sparse checking.

---

Are you sure this patch even applies? What tree did you make it against?

I ask because this fragment:

drivers/staging/gasket/apex_driver.c line 13

Quoted 3 lines
drivers/staging/gasket/apex_driver.c +0 -0
@@ -12,6 +12,7 @@
  * GNU General Public License for more details.
  */

is no longer in the tree.

thanks,

greg k-h

A message that is not in the archive (sent off-list or not yet received): CAO3upobyf9JpKLSoSPeVfNGOh6R0cud=hT+bhA=U2kNkGEJ-aw@mail.gmail.com

#39 lore raw

Re: [PATCH 12/18] staging: gasket: annotate ioctl arg with __user

On Sat, Jul 14, 2018 at 12:12:38AM -0700, Todd Poynor wrote:

HTML email gets rejected by the mailing lists :(

Please fix your email client to not do that.

Quoted 17 lines

On Fri, Jul 13, 2018 at 11:35 PM, Greg Kroah-Hartman < [email protected]> wrote:

On Fri, Jul 13, 2018 at 10:58:10PM -0700, Todd Poynor wrote:
> From: Todd Poynor <[email protected]>
>
> For sparse checking.
>
> Reported-by: Dmitry Torokhov <[email protected]> > Signed-off-by: Zhongze Hu <[email protected]> > Signed-off-by: Todd Poynor <[email protected]> > ---

Are you sure this patch even applies?  What tree did you make it against?

Quoted 5 lines

I'm working in linux-next at tag next-20180713, which is also tagged master at this writing.

I think I'm following http://www.kroah.com/log/linux/linux-staging-update.html,
but if I'm off in the weeds do clue me in, thanks.

That's something I wrote 9 years ago :)

Anyway, yes, linux-next is usually the right place to work off of, but it lags my staging.git tree usually by at least a day, maybe two, depending on when I push and when -next picks my tree up. Also there is the lag of the weekend when -next is not being generated.

Normally that is fine, except for when working on code that is "fast moving". This driver has had lots of submissions against it lately as there are loads of "low hanging fruit" that people easily see to fix up. Look at the mailing list archives for this week alone for all of those patches.

When working on something that lots of other people are working on, pay attention to the patches on the mailing list and work off of my staging.git tree, usually the staging-next branch, but sometimes staging-testing is needed if lots of patches are being sent rapidly.

Quoted 11 lines

I ask because this fragment:

> @@ -12,6 +12,7 @@
>   * GNU General Public License for more details. >   */

is no longer in the tree.

Sorry, which tree?  I've made sure my linux-next is up-to-date and I still see this code.

Look at my staging-next branch in staging.git. It's long-gone :)

thanks,

greg k-h

#40 lore raw

Re: [PATCH 12/18] staging: gasket: annotate ioctl arg with __user

Quoted 4 lines
Quoted 2 lines

I think I'm following http://www.kroah.com/log/linux/linux-staging-update.html,
but if I'm off in the weeds do clue me in, thanks.

That's something I wrote 9 years ago :)

Still the top search result for how to work in staging, suggest an update. :)

#41 lore raw

Re: [PATCH 12/18] staging: gasket: annotate ioctl arg with __user

Quoted 6 lines · Todd Poynor
Quoted 4 lines
Quoted 2 lines

I think I'm following http://www.kroah.com/log/linux/linux-staging-update.html,
but if I'm off in the weeds do clue me in, thanks.

That's something I wrote 9 years ago :)

Still the top search result for how to work in staging, suggest an update. :)

Do all subsystems have such descriptions? No, working in staging is no different from working in any other kernel subsystem. If you want to find the location of the latest code, look in the MAINTAINERS file for the git tree.

There's nothing "special" about drivers/staging, except the quality of the code...

greg k-h

#42 lore raw

Re: [PATCH 12/18] staging: gasket: annotate ioctl arg with __user

Yes you're right, I see the entry now.

Thanks -- Todd

Quoted 17 lines · Greg Kroah-Hartman
Quoted 6 lines · Todd Poynor
Quoted 4 lines
Quoted 2 lines

I think I'm following http://www.kroah.com/log/linux/linux-staging-update.html,
but if I'm off in the weeds do clue me in, thanks.

That's something I wrote 9 years ago :)

Still the top search result for how to work in staging, suggest an update. :)

Do all subsystems have such descriptions? No, working in staging is no different from working in any other kernel subsystem. If you want to find the location of the latest code, look in the MAINTAINERS file for the git tree.

There's nothing "special" about drivers/staging, except the quality of the code...

greg k-h

Signature · 1 line
Todd
#43 lore raw

PATCH 13/18 staging: gasket: gasket_enable_dev fixups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Remove unnecessary variable.

Bail out if no physical device.

---
 drivers/staging/gasket/gasket_core.c | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)
drivers/staging/gasket/gasket_core.c +6 -13
index 65aa7cf454fb..f7d8f66e8746 100644
@@ -915,8 +915,6 @@ static int gasket_enable_dev(
 {
 	int tbl_idx;
 	int ret;
-	bool has_dma_ops;
-	struct device *ddev;
 	const struct gasket_driver_desc *driver_desc =
 		internal_desc->driver_desc;
 
@@ -934,26 +932,21 @@ static int gasket_enable_dev(
 		return ret;
 	}
 
-	has_dma_ops = true;
+	if (!gasket_dev->pci_dev) {
+		gasket_log_error(gasket_dev,
+				 "%s: no physical device", __func__);
+		return -ENODEV;
+	}
 
 	for (tbl_idx = 0; tbl_idx < driver_desc->num_page_tables; tbl_idx++) {
 		gasket_log_debug(
 			gasket_dev, "Initializing page table %d.", tbl_idx);
-		if (gasket_dev->pci_dev) {
-			ddev = &gasket_dev->pci_dev->dev;
-		} else {
-			gasket_log_error(
-				gasket_dev,
-				"gasket_enable_dev with no physical device!!");
-			WARN_ON(1);
-			ddev = NULL;
-		}
 		ret = gasket_page_table_init(
 			&gasket_dev->page_table[tbl_idx],
 			&gasket_dev->bar_data[
 				driver_desc->page_table_bar_index],
 			&driver_desc->page_table_configs[tbl_idx],
-			ddev, gasket_dev->pci_dev, has_dma_ops);
+			&gasket_dev->pci_dev->dev, gasket_dev->pci_dev, true);
 		if (ret) {
 			gasket_log_error(
 				gasket_dev,
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#44 lore raw

Re: [PATCH 13/18] staging: gasket: gasket_enable_dev fixups

Quoted 25 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Remove unnecessary variable.

Bail out if no physical device.

---
 drivers/staging/gasket/gasket_core.c | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)
drivers/staging/gasket/gasket_core.c +0 -2
index 65aa7cf454fb..f7d8f66e8746 100644
@@ -915,8 +915,6 @@ static int gasket_enable_dev(
 {
 	int tbl_idx;
 	int ret;
-	bool has_dma_ops;
-	struct device *ddev;
 	const struct gasket_driver_desc *driver_desc =
 		internal_desc->driver_desc;
 

drivers/staging/gasket/gasket_core.c static int gasket_enable_dev( · line 939

Quoted 9 lines
drivers/staging/gasket/gasket_core.c +5 -1
@@ -934,26 +932,21 @@ static int gasket_enable_dev(
 		return ret;
 	}
 
-	has_dma_ops = true;
+	if (!gasket_dev->pci_dev) {
+		gasket_log_error(gasket_dev,
+				 "%s: no physical device", __func__);
+		return -ENODEV;
+	}

How can this ever be true? I don't see where this can be removed to enable this error check to be hit. What am I missing?

thanks,

greg k-h

#45 lore raw

Re: [PATCH 13/18] staging: gasket: gasket_enable_dev fixups

Quoted 29 lines · Greg Kroah-Hartman
Quoted 28 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Remove unnecessary variable.

Bail out if no physical device.

---
 drivers/staging/gasket/gasket_core.c | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)
drivers/staging/gasket/gasket_core.c +0 -2
index 65aa7cf454fb..f7d8f66e8746 100644
@@ -915,8 +915,6 @@ static int gasket_enable_dev(
 {
      int tbl_idx;
      int ret;
-     bool has_dma_ops;
-     struct device *ddev;
      const struct gasket_driver_desc *driver_desc =
              internal_desc->driver_desc;
 
@@ -934,26 +932,21 @@ static int gasket_enable_dev(
              return ret;
      }
 
Quoted 9 lines

drivers/staging/gasket/gasket_core.c static int gasket_enable_dev( · line 939

Quoted 6 lines
drivers/staging/gasket/gasket_core.c +5 -1
-     has_dma_ops = true;
+     if (!gasket_dev->pci_dev) {
+             gasket_log_error(gasket_dev,
+                              "%s: no physical device", __func__);
+             return -ENODEV;
+     }

How can this ever be true? I don't see where this can be removed to enable this error check to be hit. What am I missing?

Agree, I'll split this into two patches for the two things it does, and just remove the unnecessary check. Thanks,

Quoted 4 lines

thanks,

greg k-h

Signature · 1 line
Todd
#46 lore raw

PATCH 14/18 staging: gasket: fix class create bug handling

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

class_create() never returns NULL, and this driver should never return PTR_ERR(NULL) anyway.

---
 drivers/staging/gasket/gasket_core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/gasket/gasket_core.c +1 -1
index f7d8f66e8746..0ef37667e0f1 100644
@@ -341,7 +341,7 @@ int gasket_register_device(const struct gasket_driver_desc *driver_desc)
 	internal->class =
 		class_create(driver_desc->module, driver_desc->name);
 
-	if (IS_ERR_OR_NULL(internal->class)) {
+	if (IS_ERR(internal->class)) {
 		gasket_nodev_error("Cannot register %s class [ret=%ld]",
 				   driver_desc->name, PTR_ERR(internal->class));
 		ret = PTR_ERR(internal->class);
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
Reviewed-by on v1 14/18
#47 lore raw

Re: [PATCH 14/18] staging: gasket: fix class create bug handling

Quoted 9 lines · Todd Poynor

From: Todd Poynor <[email protected]>

class_create() never returns NULL, and this driver should never return PTR_ERR(NULL) anyway.

Quoted 20 lines
---
 drivers/staging/gasket/gasket_core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/gasket/gasket_core.c +1 -2
index f7d8f66e8746..0ef37667e0f1 100644
@@ -341,7 +341,7 @@ int gasket_register_device(const struct gasket_driver_desc *driver_desc)
        internal->class =
                class_create(driver_desc->module, driver_desc->name);
 
-       if (IS_ERR_OR_NULL(internal->class)) {
+       if (IS_ERR(internal->class)) {
                gasket_nodev_error("Cannot register %s class [ret=%ld]",
                                   driver_desc->name, PTR_ERR(internal->class));
                ret = PTR_ERR(internal->class);
--

2.18.0.203.gfac676dfb9-goog

#48 lore raw

PATCH 15/18 staging: gasket: remove unnecessary code in coherent allocator

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Remove extraneous statement in gasket_config_coherent_allocator()

---
 drivers/staging/gasket/gasket_ioctl.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/staging/gasket/gasket_ioctl.c +1 -3
index 0c1e0723bef1..fcf58f44efed 100644
@@ -427,10 +427,8 @@ static int gasket_config_coherent_allocator(
 	if (ibuf.page_table_index >= gasket_dev->num_page_tables)
 		return -EFAULT;
 
-	if (ibuf.size > PAGE_SIZE * MAX_NUM_COHERENT_PAGES) {
-		ibuf.size = PAGE_SIZE * MAX_NUM_COHERENT_PAGES;
+	if (ibuf.size > PAGE_SIZE * MAX_NUM_COHERENT_PAGES)
 		return -ENOMEM;
-	}
 
 	if (ibuf.enable == 0) {
 		ret = gasket_free_coherent_memory(
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#49 lore raw

PATCH 16/18 staging: gasket: gasket core error handling fixups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Minor fixups to error codes and error handling in gasket core.

No device reset callback registered is not an error condition.

---
 drivers/staging/gasket/gasket_core.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)
drivers/staging/gasket/gasket_core.c +5 -8
index 0ef37667e0f1..65780d4dffbf 100644
@@ -814,7 +814,7 @@ static int gasket_map_pci_bar(struct gasket_dev *gasket_dev, int bar_num)
 			gasket_dev,
 			"Cannot get BAR %d memory region %p",
 			bar_num, &gasket_dev->pci_dev->resource[bar_num]);
-		return -EINVAL;
+		return -EBUSY;
 	}
 
 	gasket_dev->bar_data[bar_num].virt_base =
@@ -1655,7 +1655,7 @@ static int gasket_mmap(struct file *filp, struct vm_area_struct *vma)
 			"0x%lx",
 			raw_offset);
 		trace_gasket_mmap_exit(bar_index);
-		return bar_index;
+		return -EINVAL;
 	}
 
 	vma->vm_private_data = gasket_dev;
@@ -1865,11 +1865,8 @@ int gasket_reset_nolock(struct gasket_dev *gasket_dev, uint reset_type)
 	const struct gasket_driver_desc *driver_desc;
 
 	driver_desc = gasket_dev->internal_desc->driver_desc;
-	if (!driver_desc->device_reset_cb) {
-		gasket_log_error(
-			gasket_dev, "No device reset callback was registered.");
-		return -EINVAL;
-	}
+	if (!driver_desc->device_reset_cb)
+		return 0;
 
 	/* Perform a device reset of the requested type. */
 	ret = driver_desc->device_reset_cb(gasket_dev, reset_type);
@@ -1894,7 +1891,7 @@ int gasket_reset_nolock(struct gasket_dev *gasket_dev, uint reset_type)
 	gasket_dev->status = gasket_get_hw_status(gasket_dev);
 	if (gasket_dev->status == GASKET_STATUS_DEAD) {
 		gasket_log_error(gasket_dev, "Device reported as dead.");
-		return -EINVAL;
+		return -EIO;
 	}
 
 	return 0;
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#50 lore raw

PATCH 17/18 staging: gasket: don't release coherent mappings

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

coherent address mappings aren't backed by a struct page, don't need to be released, and don't count as an active page in the page table bookkeeping.

---
 drivers/staging/gasket/gasket_page_table.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/gasket/gasket_page_table.c +1 -1
index d7600d8e385f..9a4a81c010f9 100644
@@ -1248,7 +1248,7 @@ static void gasket_perform_unmapping(
 				dma_unmap_page(pg_tbl->device, ptes[i].dma_addr,
 					       PAGE_SIZE, DMA_FROM_DEVICE);
 			}
-			if (gasket_release_page(ptes[i].page))
+			if (ptes[i].page && gasket_release_page(ptes[i].page))
 				--pg_tbl->num_active_pages;
 		}
 		ptes[i].status = PTE_FREE;
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#51 lore raw

PATCH 18/18 staging: gasket: various cleanups

In reply to Todd Poynor

From: Todd Poynor <[email protected]>

Cleanups to error codes, code style, error conditions, etc.

---
 drivers/staging/gasket/apex.h              |  7 ++-
 drivers/staging/gasket/apex_driver.c       | 50 ++++++++---------
 drivers/staging/gasket/gasket_core.c       | 10 ++--
 drivers/staging/gasket/gasket_core.h       |  3 +-
 drivers/staging/gasket/gasket_ioctl.c      |  9 +++
 drivers/staging/gasket/gasket_page_table.c | 65 +++++++++++-----------
 drivers/staging/gasket/gasket_page_table.h |  8 +--
 7 files changed, 80 insertions(+), 72 deletions(-)
drivers/staging/gasket/apex.h +4 -3
index f2600aa05191..8c2f004d4427 100644
@@ -30,9 +30,10 @@
 
 #define APEX_EXTENDED_SHIFT 63 /* Extended address bit position. */
 
-/* Addresses are 2^3=8 bytes each. */
-/* page in second level page table */
-/* holds APEX_PAGE_SIZE/8 addresses  */
+/*
+ * Addresses are 2^3=8 bytes each. Page in second level page table holds
+ * APEX_PAGE_SIZE/8 addresses.
+ */
 #define APEX_ADDR_SHIFT 3
 #define APEX_LEVEL_SHIFT (APEX_PAGE_SHIFT - APEX_ADDR_SHIFT)
 #define APEX_LEVEL_SIZE BIT(APEX_LEVEL_SHIFT)
drivers/staging/gasket/apex_driver.c +24 -26

73 lines not on this page · raw message

drivers/staging/gasket/gasket_core.c +5 -5
index 65780d4dffbf..fc89654e562d 100644
@@ -1619,10 +1619,10 @@ static int gasket_mmap(struct file *filp, struct vm_area_struct *vma)
 	}
 	driver_desc = gasket_dev->internal_desc->driver_desc;
 
-	if (vma->vm_start & (PAGE_SIZE - 1)) {
+	if (vma->vm_start & ~PAGE_MASK) {
 		gasket_log_error(
-			gasket_dev, "Base address not page-aligned: 0x%p\n",
-			(void *)vma->vm_start);
+			gasket_dev, "Base address not page-aligned: %lx\n",
+			vma->vm_start);
 		trace_gasket_mmap_exit(-EINVAL);
 		return -EINVAL;
 	}
@@ -1917,7 +1917,7 @@ static ssize_t gasket_write_mappable_regions(
 	if (bar_desc.permissions == GASKET_NOMAP)
 		return 0;
 	for (i = 0;
-	     (i < bar_desc.num_mappable_regions) && (total_written < PAGE_SIZE);
+	     i < bar_desc.num_mappable_regions && total_written < PAGE_SIZE;
 	     i++) {
 		min_addr = bar_desc.mappable_regions[i].start -
 			   driver_desc->legacy_mmap_address_offset;
@@ -2081,7 +2081,7 @@ struct device *gasket_get_device(struct gasket_dev *dev)
  *
  * Description: Busy waits for a specific combination of bits to be set
  * on a Gasket register.
- **/
+ */
 int gasket_wait_sync(
 	struct gasket_dev *gasket_dev, int bar, u64 offset, u64 mask, u64 val,
 	u64 timeout_ns)
drivers/staging/gasket/gasket_core.h +2 -1
index 177b239e2c50..ab3d69ee6faa 100644
@@ -61,7 +61,8 @@ enum gasket_interrupt_type {
 	PLATFORM_WIRE = 2,
 };
 
-/* Used to describe a Gasket interrupt. Contains an interrupt index, a register,
+/*
+ * Used to describe a Gasket interrupt. Contains an interrupt index, a register,
  * and packing data for that interrupt. The register and packing data
  * fields are relevant only for PCI_MSIX interrupt type and can be
  * set to 0 for everything else.
drivers/staging/gasket/gasket_ioctl.c +9 -0
index fcf58f44efed..e6b27f5c067c 100644
@@ -361,6 +361,9 @@ static int gasket_map_buffers(struct gasket_dev *gasket_dev,
 		ibuf.page_table_index, ibuf.size, ibuf.host_address,
 		ibuf.device_address);
 
+	if (ibuf.size < PAGE_SIZE)
+		return -EINVAL;
+
 	if (ibuf.page_table_index >= gasket_dev->num_page_tables)
 		return -EFAULT;
 
@@ -424,12 +427,18 @@ static int gasket_config_coherent_allocator(
 	trace_gasket_ioctl_config_coherent_allocator(
 		ibuf.enable, ibuf.size, ibuf.dma_address);
 
+	if (ibuf.size < PAGE_SIZE)
+		return -EINVAL;
+
 	if (ibuf.page_table_index >= gasket_dev->num_page_tables)
 		return -EFAULT;
 
 	if (ibuf.size > PAGE_SIZE * MAX_NUM_COHERENT_PAGES)
 		return -ENOMEM;
 
+	if (gasket_dev->page_table[ibuf.page_table_index] == 0)
+		return -EINVAL;
+
 	if (ibuf.enable == 0) {
 		ret = gasket_free_coherent_memory(
 			gasket_dev, ibuf.size, ibuf.dma_address,
drivers/staging/gasket/gasket_page_table.c +32 -33

207 lines not on this page · raw message

drivers/staging/gasket/gasket_page_table.h +4 -4
index f2f519a3022d..9180b56984ce 100644
@@ -169,9 +169,9 @@ int gasket_page_table_lookup_page(
  * specified by both addresses and the size are valid for mapping pages into
  * device memory.
  *
- * Returns 1 if true - if the mapping is bad, 0 otherwise.
+ * Returns true if the mapping is bad, false otherwise.
  */
-int gasket_page_table_are_addrs_bad(
+bool gasket_page_table_are_addrs_bad(
 	struct gasket_page_table *page_table, ulong host_addr, ulong dev_addr,
 	ulong bytes);
 
@@ -185,9 +185,9 @@ int gasket_page_table_are_addrs_bad(
  * specified by the device address and the size is valid for mapping pages into
  * device memory.
  *
- * Returns 1 if true - if the address is bad, 0 otherwise.
+ * Returns true if the address is bad, false otherwise.
  */
-int gasket_page_table_is_dev_addr_bad(
+bool gasket_page_table_is_dev_addr_bad(
 	struct gasket_page_table *page_table, ulong dev_addr, ulong bytes);
 
 /*
Signature · 1 line
2.18.0.203.gfac676dfb9-goog
#52 lore raw

Re: [PATCH 18/18] staging: gasket: various cleanups

Quoted 3 lines · Todd Poynor

From: Todd Poynor <[email protected]>

Cleanups to error codes, code style, error conditions, etc.

You you need to break this up into "one type of logical change per patch" here for this change, sorry.

thanks,

greg k-h