Re: [ANNOUNCE] WOLK v2.3 for Kernel v2.6.4

Re: [ANNOUNCE] WOLK v2.3 for Kernel v2.6.4

[Shantanu Goel]

[-- Attachment #1: Type: text/plain, Size: 559 bytes --]

Hi Marc,

I took the patch for a spin and found a few glitches.
A patch is attached to fix them.

1. The kernel oops'ed in vt_ioctl due to driver_data
being NULL in the tty.  I copied the relevant fix from
2.6.5-rc2-mm2.

2. I did not configure the in-kernel statd daemon for
NFS.  That caused a BUG() in mon.c:nsm_monitor()
because the handle was never initialized.

3. A typo fix in ipmi/af_ipmi.c.

Thanks,
Shantanu

__________________________________
Do you Yahoo!?
Yahoo! Finance Tax Center - File online. File on time.
http://taxes.yahoo.com/filing.html

[-- Attachment #2: wolk.patch --]
[-- Type: application/octet-stream, Size: 2683 bytes --]

--- ./drivers/char/vt.c.~1~	2004-03-26 21:32:05.102510491 -0500
+++ ./drivers/char/vt.c	2004-03-26 23:04:41.443943776 -0500
@@ -2509,13 +2509,11 @@
 	if (tty && tty->count == 1) {
 		struct vt_struct *vt;
 
+		vcs_remove_devfs(tty);
 		vt = tty->driver_data;
 		if (vt)
 			vc_cons[vt->vc_num].d->vc_tty = NULL;
 		tty->driver_data = 0;
-		release_console_sem();
-		vcs_remove_devfs(tty);
-		return;
 	}
 	release_console_sem();
 }
--- ./fs/lockd/host.c.~1~	2004-03-26 21:32:14.000000000 -0500
+++ ./fs/lockd/host.c	2004-03-26 23:37:51.842001343 -0500
@@ -55,6 +55,25 @@
 }
 
 /*
+ * Allocate an NSM handle
+ */
+struct nsm_handle *
+nsm_alloc(struct sockaddr_in *sin)
+{
+	struct nsm_handle *nsm;
+
+	nsm = (struct nsm_handle *) kmalloc(sizeof(*nsm), GFP_KERNEL);
+	if (nsm == NULL)
+		return NULL;
+
+	memset(nsm, 0, sizeof(*nsm));
+	memcpy(&nsm->sm_addr, sin, sizeof(nsm->sm_addr));
+	atomic_set(&nsm->sm_count, 1);
+
+	return nsm;
+}
+
+/*
  * Common host lookup routine for server & client
  */
 struct nlm_host *
@@ -143,10 +162,10 @@
 			}
 		}
 	}
+#endif
 
 	if (host->h_nsmhandle == NULL)
 		host->h_nsmhandle = nsm_alloc(&host->h_addr);
-#endif
 
 	if (++nrhosts > NLM_HOST_MAX)
 		next_gc = 0;
--- ./fs/lockd/mon.c.~1~	2004-03-26 21:32:14.597598271 -0500
+++ ./fs/lockd/mon.c	2004-03-26 23:45:59.915057066 -0500
@@ -98,7 +98,7 @@
 {
 	struct nsm_handle *nsm;
 	struct nsm_res	res;
-	int		status;
+	int		status = 0;
 
 	nsm = host->h_nsmhandle;
 	host->h_nsmhandle = NULL;
@@ -112,9 +112,8 @@
 		if (status < 0)
 			printk(KERN_NOTICE "lockd: cannot unmonitor %s\n",
 				       	host->h_name);
-		else
-			nsm->sm_monitored = 0;
 	}
+	kfree(nsm);
 	return status;
 }
 
--- ./fs/lockd/statd.c.~1~	2004-03-26 21:32:14.598598175 -0500
+++ ./fs/lockd/statd.c	2004-03-26 23:37:20.990970192 -0500
@@ -189,25 +189,6 @@
 }
 
 /*
- * Allocate an NSM handle
- */
-struct nsm_handle *
-nsm_alloc(struct sockaddr_in *sin)
-{
-	struct nsm_handle *nsm;
-
-	nsm = (struct nsm_handle *) kmalloc(sizeof(*nsm), GFP_KERNEL);
-	if (nsm == NULL)
-		return NULL;
-
-	memset(nsm, 0, sizeof(*nsm));
-	memcpy(&nsm->sm_addr, sin, sizeof(nsm->sm_addr));
-	atomic_set(&nsm->sm_count, 1);
-
-	return nsm;
-}
-
-/*
  * Set up monitoring of a remote host
  * Note we hold the semaphore for the host table while
  * we're here.
--- ./net/ipmi/af_ipmi.c.~1~	2004-03-26 21:32:27.148392509 -0500
+++ ./net/ipmi/af_ipmi.c	2004-03-26 22:03:34.077745113 -0500
@@ -515,7 +515,7 @@
 {
 	struct ipmi_sock *i;
 
-	if (atomic_read(&ipmi_nr_socks) >= 2*files_stat.max_files)
+	if (atomic_read(&ipmi_nr_socks) >= 2*files_stat.max_files_hard)
 		return NULL;
 
 	i = (struct ipmi_sock *)sk_alloc(PF_IPMI, GFP_KERNEL,

Re[2]: [ANNOUNCE] WOLK v2.3 for Kernel v2.6.4

[Maciej Soltysiak]

Just a thought...

If it is supposed to be _working_ overload linux kernel, so
maybe Marc should drop the -mm patches as they contain testing
code, which is more prone to bugs than vanilla kernels.

Besides that, WOLK is great with all the stuff you might need in
there, thanks.

Regards,
Maciej

Re: [ANNOUNCE] WOLK v2.3 for Kernel v2.6.4

[Marc-Christian Petersen]

On Saturday 27 March 2004 06:21, Shantanu Goel wrote:

Hi Shantanu,

> 1. The kernel oops'ed in vt_ioctl due to driver_data
> being NULL in the tty.  I copied the relevant fix from
> 2.6.5-rc2-mm2.
> 2. I did not configure the in-kernel statd daemon for
> NFS.  That caused a BUG() in mon.c:nsm_monitor()
> because the handle was never initialized.
> 3. A typo fix in ipmi/af_ipmi.c.

thank you :) ... Applied.

ciao, Marc