[PATCH 2/7] vgaarb: Fix a few checkpatch errors and warnings

From: Thierry Reding
Date: Fri May 08 2015 - 06:51:28 EST


From: Thierry Reding <treding@xxxxxxxxxx>

Wrap overly long lines (offending lines were mostly comments, so trivial
to fix up) and a number of other coding style issues pointed out by the
checkpatch tool.

Signed-off-by: Thierry Reding <treding@xxxxxxxxxx>
---
drivers/gpu/vga/vgaarb.c | 102 ++++++++++++++++++++++++++++-------------------
1 file changed, 60 insertions(+), 42 deletions(-)

diff --git a/drivers/gpu/vga/vgaarb.c b/drivers/gpu/vga/vgaarb.c
index 5f1bde5efc7f..a0b433456107 100644
--- a/drivers/gpu/vga/vgaarb.c
+++ b/drivers/gpu/vga/vgaarb.c
@@ -136,7 +136,6 @@ struct pci_dev *vga_default_device(void)
{
return vga_default;
}
-
EXPORT_SYMBOL_GPL(vga_default_device);

void vga_set_default_device(struct pci_dev *pdev)
@@ -300,9 +299,9 @@ enable_them:

pci_set_vga_state(vgadev->pdev, true, pci_bits, flags);

- if (!vgadev->bridge_has_one_vga) {
+ if (!vgadev->bridge_has_one_vga)
vga_irq_set_state(vgadev, true);
- }
+
vgadev->owns |= wants;
lock_them:
vgadev->locks |= (rsrc & VGA_RSRC_LEGACY_MASK);
@@ -454,15 +453,15 @@ bail:
}
EXPORT_SYMBOL(vga_put);

-/* Rules for using a bridge to control a VGA descendant decoding:
- if a bridge has only one VGA descendant then it can be used
- to control the VGA routing for that device.
- It should always use the bridge closest to the device to control it.
- If a bridge has a direct VGA descendant, but also have a sub-bridge
- VGA descendant then we cannot use that bridge to control the direct VGA descendant.
- So for every device we register, we need to iterate all its parent bridges
- so we can invalidate any devices using them properly.
-*/
+/*
+ * Rules for using a bridge to control a VGA descendant decoding: if a bridge
+ * has only one VGA descendant then it can be used to control the VGA routing
+ * for that device. It should always use the bridge closest to the device to
+ * control it. If a bridge has a direct VGA descendant, but also have a sub-
+ * bridge VGA descendant then we cannot use that bridge to control the direct
+ * VGA descendant. So for every device we register, we need to iterate all
+ * its parent bridges so we can invalidate any devices using them properly.
+ */
static void vga_arbiter_check_bridge_sharing(struct vga_device *vgadev)
{
struct vga_device *same_bridge_vgadev;
@@ -486,21 +485,26 @@ static void vga_arbiter_check_bridge_sharing(struct vga_device *vgadev)

/* see if the share a bridge with this device */
if (new_bridge == bridge) {
- /* if their direct parent bridge is the same
- as any bridge of this device then it can't be used
- for that device */
+ /*
+ * If their direct parent bridge is the same
+ * as any bridge of this device then it can't
+ * be used for that device.
+ */
same_bridge_vgadev->bridge_has_one_vga = false;
}

- /* now iterate the previous devices bridge hierarchy */
- /* if the new devices parent bridge is in the other devices
- hierarchy then we can't use it to control this device */
+ /*
+ * Now iterate the previous devices bridge hierarchy.
+ * If the new devices parent bridge is in the other
+ * devices hierarchy then we can't use it to control
+ * this device
+ */
while (bus) {
bridge = bus->self;
- if (bridge) {
- if (bridge == vgadev->pdev->bus->self)
- vgadev->bridge_has_one_vga = false;
- }
+
+ if (bridge && bridge == vgadev->pdev->bus->self)
+ vgadev->bridge_has_one_vga = false;
+
bus = bus->parent;
}
}
@@ -530,9 +534,9 @@ static bool vga_arbiter_add_pci_device(struct pci_dev *pdev)
vgadev = kmalloc(sizeof(struct vga_device), GFP_KERNEL);
if (vgadev == NULL) {
pr_err("failed to allocate pci device\n");
- /* What to do on allocation failure ? For now, let's
- * just do nothing, I'm not sure there is anything saner
- * to be done
+ /*
+ * What to do on allocation failure ? For now, let's just do
+ * nothing, I'm not sure there is anything saner to be done.
*/
return false;
}
@@ -568,8 +572,8 @@ static bool vga_arbiter_add_pci_device(struct pci_dev *pdev)
bridge = bus->self;
if (bridge) {
u16 l;
- pci_read_config_word(bridge, PCI_BRIDGE_CONTROL,
- &l);
+
+ pci_read_config_word(bridge, PCI_BRIDGE_CONTROL, &l);
if (!(l & PCI_BRIDGE_CTL_VGA)) {
vgadev->owns = 0;
break;
@@ -677,7 +681,9 @@ static inline void vga_update_device_decodes(struct vga_device *vgadev,
pr_debug("decoding count now is: %d\n", vga_decode_count);
}

-static void __vga_set_legacy_decoding(struct pci_dev *pdev, unsigned int decodes, bool userspace)
+static void __vga_set_legacy_decoding(struct pci_dev *pdev,
+ unsigned int decodes,
+ bool userspace)
{
struct vga_device *vgadev;
unsigned long flags;
@@ -713,7 +719,8 @@ EXPORT_SYMBOL(vga_set_legacy_decoding);
/* call with NULL to unregister */
int vga_client_register(struct pci_dev *pdev, void *cookie,
void (*irq_set_state)(void *cookie, bool state),
- unsigned int (*set_vga_decode)(void *cookie, bool decode))
+ unsigned int (*set_vga_decode)(void *cookie,
+ bool decode))
{
int ret = -ENODEV;
struct vga_device *vgadev;
@@ -833,7 +840,7 @@ static int vga_pci_str_to_vars(char *buf, int count, unsigned int *domain,
return 1;
}

-static ssize_t vga_arb_read(struct file *file, char __user * buf,
+static ssize_t vga_arb_read(struct file *file, char __user *buf,
size_t count, loff_t *ppos)
{
struct vga_arb_private *priv = file->private_data;
@@ -900,7 +907,7 @@ done:
* TODO: To avoid parsing inside kernel and to improve the speed we may
* consider use ioctl here
*/
-static ssize_t vga_arb_write(struct file *file, const char __user * buf,
+static ssize_t vga_arb_write(struct file *file, const char __user *buf,
size_t count, loff_t *ppos)
{
struct vga_arb_private *priv = file->private_data;
@@ -1154,7 +1161,7 @@ done:
return ret_val;
}

-static unsigned int vga_arb_fpoll(struct file *file, poll_table * wait)
+static unsigned int vga_arb_fpoll(struct file *file, poll_table *wait)
{
struct vga_arb_private *priv = file->private_data;

@@ -1250,7 +1257,8 @@ static void vga_arbiter_notify_clients(void)
else
new_state = true;
if (vgadev->set_vga_decode) {
- new_decodes = vgadev->set_vga_decode(vgadev->cookie, new_state);
+ new_decodes = vgadev->set_vga_decode(vgadev->cookie,
+ new_state);
vga_update_device_decodes(vgadev, new_decodes);
}
}
@@ -1320,17 +1328,25 @@ static int __init vga_arb_device_init(void)

list_for_each_entry(vgadev, &vga_list, list) {
#if defined(CONFIG_X86) || defined(CONFIG_IA64)
- /* Override I/O based detection done by vga_arbiter_add_pci_device()
- * as it may take the wrong device (e.g. on Apple system under EFI).
+ /*
+ * Override vga_arbiter_add_pci_device()'s I/O based detection
+ * as it may take the wrong device (e.g. on Apple system under
+ * EFI).
*
- * Select the device owning the boot framebuffer if there is one.
+ * Select the device owning the boot framebuffer if there is
+ * one.
*/
- resource_size_t start, end;
+ resource_size_t start, end, limit;
+ unsigned long flags;
int i;

+ limit = screen_info.lfb_base + screen_info.lfb_size;
+
/* Does firmware framebuffer belong to us? */
for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
- if (!(pci_resource_flags(vgadev->pdev, i) & IORESOURCE_MEM))
+ flags = pci_resource_flags(vgadev->pdev, i);
+
+ if ((flags & IORESOURCE_MEM) == 0)
continue;

start = pci_resource_start(vgadev->pdev, i);
@@ -1339,9 +1355,9 @@ static int __init vga_arb_device_init(void)
if (!start || !end)
continue;

- if (screen_info.lfb_base < start ||
- (screen_info.lfb_base + screen_info.lfb_size) >= end)
+ if (screen_info.lfb_base < start || limit >= end)
continue;
+
if (!vga_default_device())
pr_info("setting as boot device: PCI:%s\n",
pci_name(vgadev->pdev));
@@ -1352,9 +1368,11 @@ static int __init vga_arb_device_init(void)
}
#endif
if (vgadev->bridge_has_one_vga)
- pr_info("bridge control possible %s\n", pci_name(vgadev->pdev));
+ pr_info("bridge control possible %s\n",
+ pci_name(vgadev->pdev));
else
- pr_info("no bridge control possible %s\n", pci_name(vgadev->pdev));
+ pr_info("no bridge control possible %s\n",
+ pci_name(vgadev->pdev));
}
return rc;
}
--
2.3.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/