[PATCH] genirq: msi: simplify the msi irq limit policy

Currently we attempt to predict how many irqs we will be able to allocate with
msi using pci_vector_resources and some complicated accounting, and then we
only allow each device as many irqs as we think are available on average.

Only the s2io driver even takes advantage of this feature all other drivers
have a fixed number of irqs they need and bail if they can't get them.

pci_vector_resources is inaccurate if anyone ever frees an irq.  The whole
implmentation is racy.  The current irq limit policy does not appear to make
sense with current drivers.  So I have simplified things.  We can revisit this
we we need a more sophisticated policy.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Rajesh Shah <rajesh.shah@intel.com>
Cc: Andi Kleen <ak@muc.de>
Cc: "Protasevich, Natalie" <Natalie.Protasevich@UNISYS.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

4 files changed, 8 insertions, 95 deletions

diff --git a/arch/i386/pci/irq.c b/arch/i386/pci/irq.c
index 4a8995c9c762..960f364cab6b 100644
--- a/arch/i386/pci/irq.c
+++ b/arch/i386/pci/irq.c
@@ -1169,33 +1169,3 @@ static int pirq_enable_irq(struct pci_dev *dev)
         }
         return 0;
 }
-
-int pci_vector_resources(int last, int nr_released)
-{
-        int count = nr_released;
-
-        int next = last;
-        int offset = (last % 8);
-
-        while (next < FIRST_SYSTEM_VECTOR) {
-                next += 8;
-#ifdef CONFIG_X86_64
-                if (next == IA32_SYSCALL_VECTOR)
-                        continue;
-#else
-                if (next == SYSCALL_VECTOR)
-                        continue;
-#endif
-                count++;
-                if (next >= FIRST_SYSTEM_VECTOR) {
-                        if (offset%8) {
-                                next = FIRST_DEVICE_VECTOR + offset;
-                                offset++;
-                                continue;
-                        }
-                        count--;
-                }
-        }
-
-        return count;
-}
diff --git a/arch/ia64/pci/pci.c b/arch/ia64/pci/pci.c
index 15c7c670da39..b30be7c48ba8 100644
--- a/arch/ia64/pci/pci.c
+++ b/arch/ia64/pci/pci.c
@@ -810,12 +810,3 @@ pcibios_prep_mwi (struct pci_dev *dev)
         }
         return rc;
 }
-
-int pci_vector_resources(int last, int nr_released)
-{
-        int count = nr_released;
-
-        count += (IA64_LAST_DEVICE_VECTOR - last);
-
-        return count;
-}
diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
index 6d9de026e141..a32e75a4417e 100644
--- a/drivers/pci/msi.c
+++ b/drivers/pci/msi.c
@@ -29,8 +29,6 @@ static kmem_cache_t* msi_cachep;
 static int pci_msi_enable = 1;
 static int last_alloc_vector;
 static int nr_released_vectors;
-static int nr_reserved_vectors = NR_HP_RESERVED_VECTORS;
-static int nr_msix_devices;
 
 #ifndef CONFIG_X86_IO_APIC
 int vector_irq[NR_VECTORS] = { [0 ... NR_VECTORS - 1] = -1};
@@ -534,11 +532,6 @@ void pci_scan_msi_device(struct pci_dev *dev)
 {
         if (!dev)
                 return;
-
-        if (pci_find_capability(dev, PCI_CAP_ID_MSIX) > 0)
-                nr_msix_devices++;
-        else if (pci_find_capability(dev, PCI_CAP_ID_MSI) > 0)
-                nr_reserved_vectors++;
 }
 
 #ifdef CONFIG_PM
@@ -869,13 +862,19 @@ static int msix_capability_init(struct pci_dev *dev,
                 attach_msi_entry(entry, vector);
         }
         if (i != nvec) {
+                int avail = i - 1;
                 i--;
                 for (; i >= 0; i--) {
                         vector = (entries + i)->vector;
                         msi_free_vector(dev, vector, 0);
                         (entries + i)->vector = 0;
                 }
-                return -EBUSY;
+                /* If we had some success report the number of irqs
+                 * we succeeded in setting up.
+                 */
+                if (avail <= 0)
+                        avail = -EBUSY;
+                return avail;
         }
         /* Set MSI-X enabled bits */
         enable_msi_mode(dev, pos, PCI_CAP_ID_MSIX);
@@ -954,14 +953,6 @@ int pci_enable_msi(struct pci_dev* dev)
                         return -EINVAL;
         }
         status = msi_capability_init(dev);
-        if (!status) {
-                if (!pos)
-                        nr_reserved_vectors--;  /* Only MSI capable */
-                else if (nr_msix_devices > 0)
-                        nr_msix_devices--;      /* Both MSI and MSI-X capable,
-                                                   but choose enabling MSI */
-        }
-
         return status;
 }
 
@@ -1070,10 +1061,9 @@ static int msi_free_vector(struct pci_dev* dev, int vector, int reassign)
  **/
 int pci_enable_msix(struct pci_dev* dev, struct msix_entry *entries, int nvec)
 {
-        int status, pos, nr_entries, free_vectors;
+        int status, pos, nr_entries;
         int i, j, temp;
         u16 control;
-        unsigned long flags;
 
         if (!entries || pci_msi_supported(dev) < 0)
                 return -EINVAL;
@@ -1112,34 +1102,7 @@ int pci_enable_msix(struct pci_dev* dev, struct msix_entry *entries, int nvec)
                 dev->irq = temp;
                 return -EINVAL;
         }
-
-        spin_lock_irqsave(&msi_lock, flags);
-        /*
-         * msi_lock is provided to ensure that enough vectors resources are
-         * available before granting.
-         */
-        free_vectors = pci_vector_resources(last_alloc_vector,
-                                nr_released_vectors);
-        /* Ensure that each MSI/MSI-X device has one vector reserved by
-           default to avoid any MSI-X driver to take all available
-           resources */
-        free_vectors -= nr_reserved_vectors;
-        /* Find the average of free vectors among MSI-X devices */
-        if (nr_msix_devices > 0)
-                free_vectors /= nr_msix_devices;
-        spin_unlock_irqrestore(&msi_lock, flags);
-
-        if (nvec > free_vectors) {
-                if (free_vectors > 0)
-                        return free_vectors;
-                else
-                        return -EBUSY;
-        }
-
         status = msix_capability_init(dev, entries, nvec);
-        if (!status && nr_msix_devices > 0)
-                nr_msix_devices--;
-
         return status;
 }
 
diff --git a/drivers/pci/msi.h b/drivers/pci/msi.h
index 3519eca50d8a..6793241f3884 100644
--- a/drivers/pci/msi.h
+++ b/drivers/pci/msi.h
@@ -8,19 +8,8 @@
 
 #include <asm/msi.h>
 
-/*
- * Assume the maximum number of hot plug slots supported by the system is about
- * ten. The worstcase is that each of these slots is hot-added with a device,
- * which has two MSI/MSI-X capable functions. To avoid any MSI-X driver, which
- * attempts to request all available vectors, NR_HP_RESERVED_VECTORS is defined
- * as below to ensure at least one message is assigned to each detected MSI/
- * MSI-X device function.
- */
-#define NR_HP_RESERVED_VECTORS  20
-
 extern int vector_irq[NR_VECTORS];
 extern void (*interrupt[NR_IRQS])(void);
-extern int pci_vector_resources(int last, int nr_released);
 
 /*
  * MSI-X Address Register