Hi Ayaz,
I know, I'm late, but:
Ayaz Abdulla wrote:
- disable_irq(dev->irq);
+
+ if (!(np->msi_flags & NV_MSI_X_ENABLED) ||
+ ((np->msi_flags & NV_MSI_X_ENABLED) &&
+ ((np->msi_flags & NV_MSI_X_VECTORS_MASK) == 0x1))) {
+ disable_irq(dev->irq);
+ } else {
+ disable_irq(np->msi_x_entry[NV_MSI_X_VECTOR_RX].vector);
+ }
What about adding a helper function?
static int nv_using_msi(dev)
{
if (...)
return 1;
return0;
}
if (nv_alloc_rx(dev)) {
spin_lock(&np->lock);
if (!np->in_shutdown)
mod_timer(&np->oom_kick, jiffies + OOM_REFILL);
spin_unlock(&np->lock);
I think this is wrong, it must be spin_lock_irq() or _irqsave().
Previously, the driver knew that the nic used only one interrupt. Thus
every codepath that runs either within the irq handler, or within
disable_irq()/enable_irq() blocks, just used spin_lock(&np->lock).
The MSI-X patch changes that: The driver has now up to 3 interrupts.
Thus the rx interrupt handler could be interrupted by the tx irq
handler. If the rx handler owns the spinlock, then this would deadlock.
@@ -1010,7 +1063,7 @@
}
if (np->tx_skbuff[skbnr]) {
- dev_kfree_skb_irq(np->tx_skbuff[skbnr]);
+ dev_kfree_skb_any(np->tx_skbuff[skbnr]);
np->tx_skbuff[skbnr] = NULL;
return 1;
} else {
What's the purpose of this change? It seems to be unrelated to the
remaining patch.
+static irqreturn_t nv_nic_irq_tx(int foo, void *data, struct pt_regs *regs)
+{
+ struct net_device *dev = (struct net_device *) data;
+ struct fe_priv *np = netdev_priv(dev);
+ u8 __iomem *base = get_hwbase(dev);
+ u32 events;
+ int i;
+
+ dprintk(KERN_DEBUG "%s: nv_nic_irq_tx\n", dev->name);
+
+ for (i=0; ; i++) {
+ events = readl(base + NvRegMSIXIrqStatus) & NVREG_IRQ_TX_ALL;
+ writel(NVREG_IRQ_TX_ALL, base + NvRegMSIXIrqStatus);
+ pci_push(base);
+ dprintk(KERN_DEBUG "%s: tx irq: %08x\n", dev->name, events);
+ if (!(events & np->irqmask))
+ break;
+
+ spin_lock(&np->lock);
Same as above - probably spin_lock_irq() is required. Theoretically, it
might be possible to split the spinlock into multiple parts, but I doubt
that this is worth teh effort.
--
Manfred
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
