Merge pull request #269 from rjarry/blacklist

Affinity mapping fixes
Irqbalance · Jul 13, 2023 · 8b23bda · 8b23bda
2 parents 4efc192 + eee7917
commit 8b23bda
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 35 deletions.
diff --git a/activate.c b/activate.c
@@ -25,10 +25,12 @@
  * of interrupts to the kernel.
  */
 #include "config.h"
+#include <errno.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <stdint.h>
+#include <string.h>
 
 #include "irqbalance.h"
 
@@ -48,7 +50,7 @@ static void activate_mapping(struct irq_info *info, void *data __attribute__((un
 {
 	char buf[PATH_MAX];
 	FILE *file;
-	int ret = 0;
+	int errsave, ret;
 	cpumask_t applied_mask;
 
 	/*
@@ -60,6 +62,9 @@ static void activate_mapping(struct irq_info *info, void *data __attribute__((un
 	if (!info->assigned_obj)
 		return;
 
+	if (info->flags & IRQ_FLAG_AFFINITY_UNMANAGED)
+		return;
+
 	/* activate only online cpus, otherwise writing to procfs returns EOVERFLOW */
 	cpus_and(applied_mask, cpu_online_map, info->assigned_obj->mask);
 
@@ -72,17 +77,37 @@ static void activate_mapping(struct irq_info *info, void *data __attribute__((un
 	sprintf(buf, "/proc/irq/%i/smp_affinity", info->irq);
 	file = fopen(buf, "w");
 	if (!file)
-		return;
+		goto error;
 
 	cpumask_scnprintf(buf, PATH_MAX, applied_mask);
 	ret = fprintf(file, "%s", buf);
-	if (ret < 0) {
-		log(TO_ALL, LOG_WARNING, "cannot change irq %i's affinity, add it to banned list", info->irq);
-		add_banned_irq(info->irq);
-		remove_one_irq_from_db(info->irq);
+	errsave = errno;
+	if (fclose(file)) {
+		errsave = errno;
+		goto error;
 	}
-	fclose(file);
+	if (ret < 0)
+		goto error;
 	info->moved = 0; /*migration is done*/
+error:
+	log(TO_ALL, LOG_WARNING,
+		"Cannot change IRQ %i affinity: %s\n",
+		info->irq, strerror(errsave));
+	switch (errsave) {
+	case ENOSPC: /* Specified CPU APIC is full. */
+	case EAGAIN: /* Interrupted by signal. */
+	case EBUSY: /* Affinity change already in progress. */
+	case EINVAL: /* IRQ would be bound to no CPU. */
+	case ERANGE: /* CPU in mask is offline. */
+	case ENOMEM: /* Kernel cannot allocate CPU mask. */
+		/* Do not blacklist the IRQ on transient errors. */
+		break;
+	default:
+		/* Any other error is considered permanent. */
+		info->flags |= IRQ_FLAG_AFFINITY_UNMANAGED;
+		log(TO_ALL, LOG_WARNING, "IRQ %i affinity is now unmanaged\n",
+			info->irq);
+	}
 }
 
 void activate_mappings(void)

diff --git a/classify.c b/classify.c
@@ -257,7 +257,7 @@ static gint compare_ints(gconstpointer a, gconstpointer b)
 	return ai->irq - bi->irq;
 }
 
-static void __add_banned_irq(int irq, GList **list)
+static void add_banned_irq(int irq, GList **list)
 {
 	struct irq_info find, *new;
 	GList *entry;
@@ -281,14 +281,9 @@ static void __add_banned_irq(int irq, GList **list)
 	return;
 }
 
-void add_banned_irq(int irq)
-{
-	__add_banned_irq(irq, &banned_irqs);
-}
-
 void add_cl_banned_irq(int irq)
 {
-	__add_banned_irq(irq, &cl_banned_irqs);
+	add_banned_irq(irq, &cl_banned_irqs);
 }
 
 gint substr_find(gconstpointer a, gconstpointer b)
@@ -392,23 +387,6 @@ static struct irq_info *add_one_irq_to_db(const char *devpath, struct irq_info *
 	return new;
 }
 
-void remove_one_irq_from_db(int irq)
-{
-	struct irq_info find, *tmp;
-	GList *entry = NULL;
-
-	find.irq = irq;
-	entry = g_list_find_custom(interrupts_db, &find, compare_ints);
-	if (!entry)
-		return;
-
-	tmp = entry->data;
-	interrupts_db = g_list_remove(interrupts_db, tmp);
-	free(tmp);
-	log(TO_CONSOLE, LOG_INFO, "IRQ %d was removed from db.\n", irq);
-	return;
-}
-
 static void parse_user_policy_key(char *buf, int irq, struct user_irq_policy *pol)
 {
 	char *key, *value, *end;
@@ -629,7 +607,7 @@ static void add_new_irq(char *path, struct irq_info *hint)
 	/* Set NULL devpath for the irq has no sysfs entries */
 	get_irq_user_policy(path, irq, &pol);
 	if ((pol.ban == 1) || check_for_irq_ban(hint, mod)) { /*FIXME*/
-		__add_banned_irq(irq, &banned_irqs);
+		add_banned_irq(irq, &banned_irqs);
 		new = get_irq_info(irq);
 	} else
 		new = add_one_irq_to_db(path, hint, &pol);

diff --git a/irqbalance.h b/irqbalance.h
@@ -109,8 +109,6 @@ extern struct irq_info *get_irq_info(int irq);
 extern void migrate_irq(GList **from, GList **to, struct irq_info *info);
 extern void free_cl_opts(void);
 extern void add_cl_banned_module(char *modname);
-extern void add_banned_irq(int irq);
-extern void remove_one_irq_from_db(int irq);
 #define irq_numa_node(irq) ((irq)->numa_node)
 
 

diff --git a/types.h b/types.h
@@ -34,7 +34,8 @@
 /*
  * IRQ Internal tracking flags
  */
-#define IRQ_FLAG_BANNED	1
+#define IRQ_FLAG_BANNED                 (1ULL << 0)
+#define IRQ_FLAG_AFFINITY_UNMANAGED     (1ULL << 1)
 
 enum obj_type_e {
 	OBJ_TYPE_CPU,