提交 1120f8b8 编写于 作者: S Stephen Hemminger 提交者: Jesse Barnes

PCI: handle long delays in VPD access

Accessing the VPD area can take a long time.  The existing
VPD access code fails consistently on my hardware. There are comments
in the SysKonnect vendor driver that it can take up to 13ms per word.

Change the access routines to:
  * use a mutex rather than spinning with IRQ's disabled and lock held
  * have a much longer timeout
  * call cond_resched while spinning
Signed-off-by: NStephen Hemminger <shemminger@vyatta.com>
Reviewed-by: NMatthew Wilcox <willy@linux.intel.com>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
上级 904d6a30
......@@ -133,39 +133,46 @@ PCI_USER_WRITE_CONFIG(dword, u32)
struct pci_vpd_pci22 {
struct pci_vpd base;
spinlock_t lock; /* controls access to hardware and the flags */
u8 cap;
struct mutex lock;
u16 flag;
bool busy;
bool flag; /* value of F bit to wait for */
u8 cap;
};
/* Wait for last operation to complete */
/*
* Wait for last operation to complete.
* This code has to spin since there is no other notification from the PCI
* hardware. Since the VPD is often implemented by serial attachment to an
* EEPROM, it may take many milliseconds to complete.
*/
static int pci_vpd_pci22_wait(struct pci_dev *dev)
{
struct pci_vpd_pci22 *vpd =
container_of(dev->vpd, struct pci_vpd_pci22, base);
u16 flag, status;
int wait;
unsigned long timeout = jiffies + HZ/20 + 2;
u16 status;
int ret;
if (!vpd->busy)
return 0;
flag = vpd->flag ? PCI_VPD_ADDR_F : 0;
wait = vpd->flag ? 10 : 1000; /* read: 100 us; write: 10 ms */
for (;;) {
ret = pci_user_read_config_word(dev,
vpd->cap + PCI_VPD_ADDR,
ret = pci_user_read_config_word(dev, vpd->cap + PCI_VPD_ADDR,
&status);
if (ret < 0)
if (ret)
return ret;
if ((status & PCI_VPD_ADDR_F) == flag) {
if ((status & PCI_VPD_ADDR_F) == vpd->flag) {
vpd->busy = false;
return 0;
}
if (wait-- == 0)
if (time_after(jiffies, timeout))
return -ETIMEDOUT;
udelay(10);
if (fatal_signal_pending(current))
return -EINTR;
if (!cond_resched())
udelay(10);
}
}
......@@ -175,7 +182,7 @@ static int pci_vpd_pci22_read(struct pci_dev *dev, int pos, int size,
struct pci_vpd_pci22 *vpd =
container_of(dev->vpd, struct pci_vpd_pci22, base);
u32 val;
int ret;
int ret = 0;
int begin, end, i;
if (pos < 0 || pos > vpd->base.len || size > vpd->base.len - pos)
......@@ -183,7 +190,9 @@ static int pci_vpd_pci22_read(struct pci_dev *dev, int pos, int size,
if (size == 0)
return 0;
spin_lock_irq(&vpd->lock);
if (mutex_lock_killable(&vpd->lock))
return -EINTR;
ret = pci_vpd_pci22_wait(dev);
if (ret < 0)
goto out;
......@@ -191,15 +200,16 @@ static int pci_vpd_pci22_read(struct pci_dev *dev, int pos, int size,
pos & ~3);
if (ret < 0)
goto out;
vpd->busy = true;
vpd->flag = 1;
vpd->flag = PCI_VPD_ADDR_F;
ret = pci_vpd_pci22_wait(dev);
if (ret < 0)
goto out;
ret = pci_user_read_config_dword(dev, vpd->cap + PCI_VPD_DATA,
&val);
out:
spin_unlock_irq(&vpd->lock);
mutex_unlock(&vpd->lock);
if (ret < 0)
return ret;
......@@ -220,7 +230,7 @@ static int pci_vpd_pci22_write(struct pci_dev *dev, int pos, int size,
struct pci_vpd_pci22 *vpd =
container_of(dev->vpd, struct pci_vpd_pci22, base);
u32 val;
int ret;
int ret = 0;
if (pos < 0 || pos > vpd->base.len || pos & 3 ||
size > vpd->base.len - pos || size < 4)
......@@ -231,7 +241,8 @@ static int pci_vpd_pci22_write(struct pci_dev *dev, int pos, int size,
val |= ((u8) *buf++) << 16;
val |= ((u32)(u8) *buf++) << 24;
spin_lock_irq(&vpd->lock);
if (mutex_lock_killable(&vpd->lock))
return -EINTR;
ret = pci_vpd_pci22_wait(dev);
if (ret < 0)
goto out;
......@@ -247,7 +258,7 @@ static int pci_vpd_pci22_write(struct pci_dev *dev, int pos, int size,
vpd->flag = 0;
ret = pci_vpd_pci22_wait(dev);
out:
spin_unlock_irq(&vpd->lock);
mutex_unlock(&vpd->lock);
if (ret < 0)
return ret;
......@@ -279,7 +290,7 @@ int pci_vpd_pci22_init(struct pci_dev *dev)
vpd->base.len = PCI_VPD_PCI22_SIZE;
vpd->base.ops = &pci_vpd_pci22_ops;
spin_lock_init(&vpd->lock);
mutex_init(&vpd->lock);
vpd->cap = cap;
vpd->busy = false;
dev->vpd = &vpd->base;
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册