PLPA Users' Mailing List Archives

From: Gleb Natapov (glebn_at_[hidden])
Date: 2006-01-10 04:18:48

• Next message: Jeff Squyres: "Re: [PLPA users] [PATCH] small fixes"
• Previous message: Jeff Squyres: "[PLPA users] 0.9a2"
• Next in thread: Jeff Squyres: "Re: [PLPA users] [PATCH] small fixes"
• Reply: Jeff Squyres: "Re: [PLPA users] [PATCH] small fixes"

I don't quite understand in what units cpusetsize should be, but if in
bytes then this patch should be applied IMHO.

Index: src/libplpa/plpa_dispatch.c
===================================================================
--- src/libplpa/plpa_dispatch.c (revision 47)
+++ src/libplpa/plpa_dispatch.c (working copy)
@@ -30,7 +30,7 @@
     switch (PLPA_NAME(api_probe)()) {
     case PLPA_PROBE_OK:
         /* This shouldn't happen, but check anyway */
- if (cpusetsize > PLPA_BITMASK_NUM_ELEMENTS) {
+ if (cpusetsize > sizeof (*cpuset)) {
             return EINVAL;
         }
         if (NULL == cpuset) {
@@ -60,7 +60,7 @@
            will accept into a temporary and use that. Again,
            efficinency is not the issue of this code -- clarity is. */
         else if (cpusetsize > PLPA_NAME(len)) {
- for (i = PLPA_NAME(len); i < cpusetsize * 8; ++i) {
+ for (i = PLPA_NAME(len) * 8; i < cpusetsize * 8; ++i) {
                 if (PLPA_CPU_ISSET(i, cpuset)) {
                     return EINVAL;
                 }
@@ -68,7 +68,7 @@
             /* No upper-level bits are set, so now copy over the bits
                that the kernel will look at */
             memset(&tmp, 0, sizeof(tmp));
- for (i = 0; i < PLPA_NAME(len); ++i) {
+ for (i = 0; i < PLPA_NAME(len) * 8; ++i) {
                 if (PLPA_CPU_ISSET(i, cpuset)) {
                     PLPA_CPU_SET(i, &tmp);
                 }

--
			Gleb.

• Next message: Jeff Squyres: "Re: [PLPA users] [PATCH] small fixes"
• Previous message: Jeff Squyres: "[PLPA users] 0.9a2"
• Next in thread: Jeff Squyres: "Re: [PLPA users] [PATCH] small fixes"
• Reply: Jeff Squyres: "Re: [PLPA users] [PATCH] small fixes"