Skip to contentSkip to conversation

2.6.14.7 -stable review

PATCH In review Greg KH 10 participants 26 messages Jan 28, 2006 – Feb 10, 2006

Subsystems (3)
lore
Share & tools
lore

Greg KH

author
#0 lore raw

cover 2.6.14.7 -stable review

This is the start of the stable review cycle for the 2.6.14.7 release. There are 6 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let us know. If anyone is a maintainer of the proper subsystem, and wants to add a signed-off-by: line to the patch, please respond with it.

These patches are sent out with a number of different people on the Cc:
line. If you wish to be a reviewer, please email [email protected] to add your name to the list. If you want to be off the reviewer list, also email us.

Responses should be made by Monday, January 30, 00:00:00 UTC. Anything received after that time, might be too late.

thanks,

the -stable release team

author
#1 lore raw

PATCH 1/6 setting irq affinity is broken in ia32 with MSI enabled

-stable review patch. If anyone has any objections, please let us know.

------------------

From: Shaohua Li <[email protected]>

Setting irq affinity stops working when MSI is enabled. With MSI, move_irq is empty, so we can't change irq affinity. It appears a typo in Ashok's original commit for this issue. X86_64 actually is using move_native_irq.

---
 arch/i386/kernel/io_apic.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/i386/kernel/io_apic.c +2 -2
@@ -1937,7 +1937,7 @@ static void ack_edge_ioapic_vector(unsig
 {
 	int irq = vector_to_irq(vector);
 
-	move_irq(vector);
+	move_native_irq(vector);
 	ack_edge_ioapic_irq(irq);
 }
 
@@ -1952,7 +1952,7 @@ static void end_level_ioapic_vector (uns
 {
 	int irq = vector_to_irq(vector);
 
-	move_irq(vector);
+	move_native_irq(vector);
 	end_level_ioapic_irq(irq);
 }
 

--

author
#2 lore raw

PATCH 2/6 Don't match tcp/udp source/destination port for IP fragments

In reply to Greg KH

-stable review patch. If anyone has any objections, please let us know.

------------------
From: Bart De Schuymer <[email protected]>

---
 net/bridge/netfilter/ebt_ip.c |    3 +++
 1 file changed, 3 insertions(+)
net/bridge/netfilter/ebt_ip.c +3 -0
@@ -15,6 +15,7 @@
 #include <linux/netfilter_bridge/ebtables.h>
 #include <linux/netfilter_bridge/ebt_ip.h>
 #include <linux/ip.h>
+#include <net/ip.h>
 #include <linux/in.h>
 #include <linux/module.h>
 
@@ -51,6 +52,8 @@ static int ebt_filter_ip(const struct sk
 		if (!(info->bitmask & EBT_IP_DPORT) &&
 		    !(info->bitmask & EBT_IP_SPORT))
 			return EBT_MATCH;
+		if (ntohs(ih->frag_off) & IP_OFFSET)
+			return EBT_NOMATCH;
 		pptr = skb_header_pointer(skb, ih->ihl*4,
 					  sizeof(_ports), &_ports);
 		if (pptr == NULL)

--

author
#3 lore raw

PATCH 3/6 : Fix ptrace/strace

In reply to Greg KH

-stable review patch. If anyone has any objections, please let us know.

------------------
From: Richard Mortimer <[email protected]>

Don't clobber register %l0 while checking TI_SYS_NOERROR value in syscall return path. This bug was introduced by:

db7d9a4eb700be766cc9f29241483dbb1e748832

Problem narrowed down by Luis F. Ortiz and Richard Mortimer.

I tried using %l2 as suggested by Luis and that works for me.

Looking at the code I wonder if it makes sense to simplify the code a little bit. The following works for me but I'm not sure how to exercise the "NOERROR" codepath.

---
 arch/sparc64/kernel/entry.S |    7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
arch/sparc64/kernel/entry.S +2 -5
@@ -1657,13 +1657,10 @@ ret_sys_call:
 	/* Check if force_successful_syscall_return()
 	 * was invoked.
 	 */
-	ldub		[%curptr + TI_SYS_NOERROR], %l0
-	brz,pt		%l0, 1f
-	 nop
-	ba,pt		%xcc, 80f
+	ldub            [%curptr + TI_SYS_NOERROR], %l2
+	brnz,a,pn       %l2, 80f
 	 stb		%g0, [%curptr + TI_SYS_NOERROR]
 
-1:
 	cmp		%o0, -ERESTART_RESTARTBLOCK
 	bgeu,pn		%xcc, 1f
 	 andcc		%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT), %l6

--

author
#4 lore raw

PATCH 4/6 : Fix sys_fstat64() entry in 64-bit syscall table.

In reply to Greg KH

-stable review patch. If anyone has any objections, please let us know.

------------------
From: "David S. Miller" <[email protected]>

Noticed by Jakub Jelinek.

---
 arch/sparc64/kernel/systbls.S |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/sparc64/kernel/systbls.S +1 -1
@@ -98,7 +98,7 @@ sys_call_table:
 	.word sys_umount, sys_setgid, sys_getgid, sys_signal, sys_geteuid
 /*50*/	.word sys_getegid, sys_acct, sys_memory_ordering, sys_nis_syscall, sys_ioctl
 	.word sys_reboot, sys_nis_syscall, sys_symlink, sys_readlink, sys_execve
-/*60*/	.word sys_umask, sys_chroot, sys_newfstat, sys_stat64, sys_getpagesize
+/*60*/	.word sys_umask, sys_chroot, sys_newfstat, sys_fstat64, sys_getpagesize
 	.word sys_msync, sys_vfork, sys_pread64, sys_pwrite64, sys_nis_syscall
 /*70*/	.word sys_nis_syscall, sys_mmap, sys_nis_syscall, sys64_munmap, sys_mprotect
 	.word sys_madvise, sys_vhangup, sys_nis_syscall, sys_mincore, sys_getgroups

--

author
#5 lore raw

PATCH 5/6 : Fix crash in ip_nat_pptp (CVE-2006-0036)

In reply to Greg KH

-stable review patch. If anyone has any objections, please let us know.

------------------

From: Patrick McHardy <[email protected]>

When an inbound PPTP_IN_CALL_REQUEST packet is received the PPTP NAT helper uses a NULL pointer in pointer arithmentic to calculate the offset in the packet which needs to be mangled and corrupts random memory or crashes.

---
 net/ipv4/netfilter/ip_nat_helper_pptp.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv4/netfilter/ip_nat_helper_pptp.c +1 -1
@@ -313,7 +313,7 @@ pptp_inbound_pkt(struct sk_buff **pskb,
 		break;
 	case PPTP_IN_CALL_REQUEST:
 		/* only need to nat in case PAC is behind NAT box */
-		break;
+		return NF_ACCEPT;
 	case PPTP_WAN_ERROR_NOTIFY:
 		pcid = &pptpReq->wanerr.peersCallID;
 		break;

--

author
#6 lore raw

PATCH 6/6 : Fix another crash in ip_nat_pptp (CVE-2006-0037)

In reply to Greg KH

-stable review patch. If anyone has any objections, please let us know.

------------------
From: Patrick McHardy <[email protected]>

The PPTP NAT helper calculates the offset at which the packet needs to be mangled as difference between two pointers to the header. With non-linear skbs however the pointers may point to two seperate buffers on the stack and the calculation results in a wrong offset beeing used.

---
 net/ipv4/netfilter/ip_nat_helper_pptp.c |   57 +++++++++++++++-----------------
 1 file changed, 27 insertions(+), 30 deletions(-)
net/ipv4/netfilter/ip_nat_helper_pptp.c +27 -30
@@ -148,14 +148,14 @@ pptp_outbound_pkt(struct sk_buff **pskb,
 {
 	struct ip_ct_pptp_master *ct_pptp_info = &ct->help.ct_pptp_info;
 	struct ip_nat_pptp *nat_pptp_info = &ct->nat.help.nat_pptp_info;
-
-	u_int16_t msg, *cid = NULL, new_callid;
+	u_int16_t msg, new_callid;
+	unsigned int cid_off;
 
 	new_callid = htons(ct_pptp_info->pns_call_id);
 	
 	switch (msg = ntohs(ctlh->messageType)) {
 		case PPTP_OUT_CALL_REQUEST:
-			cid = &pptpReq->ocreq.callID;
+			cid_off = offsetof(union pptp_ctrl_union, ocreq.callID);
 			/* FIXME: ideally we would want to reserve a call ID
 			 * here.  current netfilter NAT core is not able to do
 			 * this :( For now we use TCP source port. This breaks
@@ -172,10 +172,10 @@ pptp_outbound_pkt(struct sk_buff **pskb,
 			ct_pptp_info->pns_call_id = ntohs(new_callid);
 			break;
 		case PPTP_IN_CALL_REPLY:
-			cid = &pptpReq->icreq.callID;
+			cid_off = offsetof(union pptp_ctrl_union, icreq.callID);
 			break;
 		case PPTP_CALL_CLEAR_REQUEST:
-			cid = &pptpReq->clrreq.callID;
+			cid_off = offsetof(union pptp_ctrl_union, clrreq.callID);
 			break;
 		default:
 			DEBUGP("unknown outbound packet 0x%04x:%s\n", msg,
@@ -197,18 +197,15 @@ pptp_outbound_pkt(struct sk_buff **pskb,
 
 	/* only OUT_CALL_REQUEST, IN_CALL_REPLY, CALL_CLEAR_REQUEST pass
 	 * down to here */
-
-	IP_NF_ASSERT(cid);
-
 	DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-		ntohs(*cid), ntohs(new_callid));
+		ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_callid));
 
 	/* mangle packet */
 	if (ip_nat_mangle_tcp_packet(pskb, ct, ctinfo,
-		(void *)cid - ((void *)ctlh - sizeof(struct pptp_pkt_hdr)),
-				 	sizeof(new_callid), 
-					(char *)&new_callid,
-				 	sizeof(new_callid)) == 0)
+	                             cid_off + sizeof(struct pptp_pkt_hdr) +
+	                             sizeof(struct PptpControlHeader),
+	                             sizeof(new_callid), (char *)&new_callid,
+	                             sizeof(new_callid)) == 0)
 		return NF_DROP;
 
 	return NF_ACCEPT;
@@ -297,7 +294,8 @@ pptp_inbound_pkt(struct sk_buff **pskb,
 		 union pptp_ctrl_union *pptpReq)
 {
 	struct ip_nat_pptp *nat_pptp_info = &ct->nat.help.nat_pptp_info;
-	u_int16_t msg, new_cid = 0, new_pcid, *pcid = NULL, *cid = NULL;
+	u_int16_t msg, new_cid = 0, new_pcid;
+	unsigned int pcid_off, cid_off = 0;
 
 	int ret = NF_ACCEPT, rv;
 
@@ -305,23 +303,23 @@ pptp_inbound_pkt(struct sk_buff **pskb,
 
 	switch (msg = ntohs(ctlh->messageType)) {
 	case PPTP_OUT_CALL_REPLY:
-		pcid = &pptpReq->ocack.peersCallID;	
-		cid = &pptpReq->ocack.callID;
+		pcid_off = offsetof(union pptp_ctrl_union, ocack.peersCallID);
+		cid_off = offsetof(union pptp_ctrl_union, ocack.callID);
 		break;
 	case PPTP_IN_CALL_CONNECT:
-		pcid = &pptpReq->iccon.peersCallID;
+		pcid_off = offsetof(union pptp_ctrl_union, iccon.peersCallID);
 		break;
 	case PPTP_IN_CALL_REQUEST:
 		/* only need to nat in case PAC is behind NAT box */
 		return NF_ACCEPT;
 	case PPTP_WAN_ERROR_NOTIFY:
-		pcid = &pptpReq->wanerr.peersCallID;
+		pcid_off = offsetof(union pptp_ctrl_union, wanerr.peersCallID);
 		break;
 	case PPTP_CALL_DISCONNECT_NOTIFY:
-		pcid = &pptpReq->disc.callID;
+		pcid_off = offsetof(union pptp_ctrl_union, disc.callID);
 		break;
 	case PPTP_SET_LINK_INFO:
-		pcid = &pptpReq->setlink.peersCallID;
+		pcid_off = offsetof(union pptp_ctrl_union, setlink.peersCallID);
 		break;
 
 	default:
@@ -343,25 +341,24 @@ pptp_inbound_pkt(struct sk_buff **pskb,
 	 * WAN_ERROR_NOTIFY, CALL_DISCONNECT_NOTIFY pass down here */
 
 	/* mangle packet */
-	IP_NF_ASSERT(pcid);
 	DEBUGP("altering peer call id from 0x%04x to 0x%04x\n",
-		ntohs(*pcid), ntohs(new_pcid));
+		ntohs(*(u_int16_t *)pptpReq + pcid_off), ntohs(new_pcid));
 	
-	rv = ip_nat_mangle_tcp_packet(pskb, ct, ctinfo, 
-				      (void *)pcid - ((void *)ctlh - sizeof(struct pptp_pkt_hdr)),
+	rv = ip_nat_mangle_tcp_packet(pskb, ct, ctinfo,
+	                              pcid_off + sizeof(struct pptp_pkt_hdr) +
+				      sizeof(struct PptpControlHeader),
 				      sizeof(new_pcid), (char *)&new_pcid, 
 				      sizeof(new_pcid));
 	if (rv != NF_ACCEPT) 
 		return rv;
 
 	if (new_cid) {
-		IP_NF_ASSERT(cid);
 		DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-			ntohs(*cid), ntohs(new_cid));
-		rv = ip_nat_mangle_tcp_packet(pskb, ct, ctinfo, 
-					      (void *)cid - ((void *)ctlh - sizeof(struct pptp_pkt_hdr)), 
-					      sizeof(new_cid),
-					      (char *)&new_cid, 
+			ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_cid));
+		rv = ip_nat_mangle_tcp_packet(pskb, ct, ctinfo,
+		                              cid_off + sizeof(struct pptp_pkt_hdr) +
+					      sizeof(struct PptpControlHeader),
+					      sizeof(new_cid), (char *)&new_cid,
 					      sizeof(new_cid));
 		if (rv != NF_ACCEPT)
 			return rv;

--

#7 lore raw

Re: [patch 6/6] [NETFILTER]: Fix another crash in ip_nat_pptp (CVE-2006-0037)

net/ipv4/netfilter/ip_nat_helper_pptp.c pptp_outbound_pkt(struct sk_buff **pskb, · line 201

Quoted 3 lines · Greg KH
net/ipv4/netfilter/ip_nat_helper_pptp.c +1 -1
 	DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-		ntohs(*cid), ntohs(new_callid));
+		ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_callid));

Note that this fix introduces another bug in the above use DEBUGP statement, as there is (u_int16_t *) ptr arithmetic used, whereas cid_off is a byte offset.

A fix for that was send a few weeks ago on netfilter-devel.

/holger

#8 lore raw

Re: [stable] Re: [patch 6/6] [NETFILTER]: Fix another crash in ip_nat_pptp (CVE-2006-0037)

Quoted 11 lines · Holger Eitzenberger

net/ipv4/netfilter/ip_nat_helper_pptp.c pptp_outbound_pkt(struct sk_buff **pskb, · line 201

Quoted 3 lines · Greg KH
net/ipv4/netfilter/ip_nat_helper_pptp.c +1 -1
 	DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-		ntohs(*cid), ntohs(new_callid));
+		ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_callid));

Note that this fix introduces another bug in the above use DEBUGP statement, as there is (u_int16_t *) ptr arithmetic used, whereas cid_off is a byte offset.

A fix for that was send a few weeks ago on netfilter-devel.

Great, care to forward it to [email protected] so we can get it in the next release?

thanks,

greg k-h

#9 lore raw

Re: [stable] Re: [patch 6/6] [NETFILTER]: Fix another crash in ip_nat_pptp (CVE-2006-0037)

Quoted 12 lines · Greg KH
Quoted 10 lines · Holger Eitzenberger

net/ipv4/netfilter/ip_nat_helper_pptp.c pptp_outbound_pkt(struct sk_buff **pskb, · line 201

Quoted 3 lines · Greg KH
net/ipv4/netfilter/ip_nat_helper_pptp.c +1 -1
 	DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-		ntohs(*cid), ntohs(new_callid));
+		ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_callid));

Note that this fix introduces another bug in the above use DEBUGP statement, as there is (u_int16_t *) ptr arithmetic used, whereas cid_off is a byte offset.

Quoted 5 lines
Quoted 1 line

A fix for that was send a few weeks ago on netfilter-devel.

Great, care to forward it to [email protected] so we can get it in the next release?

I have a copy of the patch and I'll cc stable@ on it. Although afaik this bug just causes wrong debug info to come out, so I don't think it's terribly important (?)

#10 lore raw

Re: [stable] Re: [patch 6/6] [NETFILTER]: Fix another crash in ip_nat_pptp (CVE-2006-0037)

Quoted 15 lines · Andrew Morton
Quoted 14 lines · Greg KH
Quoted 11 lines · Holger Eitzenberger

net/ipv4/netfilter/ip_nat_helper_pptp.c pptp_outbound_pkt(struct sk_buff **pskb, · line 201

Quoted 3 lines · Greg KH
net/ipv4/netfilter/ip_nat_helper_pptp.c +1 -1
 	DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-		ntohs(*cid), ntohs(new_callid));
+		ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_callid));

Note that this fix introduces another bug in the above use DEBUGP statement, as there is (u_int16_t *) ptr arithmetic used, whereas cid_off is a byte offset.

A fix for that was send a few weeks ago on netfilter-devel.

Quoted 7 lines
Quoted 3 lines

Great, care to forward it to [email protected] so we can get it in the next release?

I have a copy of the patch and I'll cc stable@ on it. Although afaik this bug just causes wrong debug info to come out, so I don't think it's terribly important (?)

If that's the only problem with it, no it's not worth adding to -stable.

thanks though.

greg k-h

#11 lore raw

Re: [stable] Re: [patch 6/6] [NETFILTER]: Fix another crash in ip_nat_pptp (CVE-2006-0037)

Quoted 20 lines · Greg KH
Quoted 19 lines · Andrew Morton
Quoted 17 lines · Greg KH
Quoted 11 lines · Holger Eitzenberger

net/ipv4/netfilter/ip_nat_helper_pptp.c pptp_outbound_pkt(struct sk_buff **pskb, · line 201

Quoted 3 lines · Greg KH
net/ipv4/netfilter/ip_nat_helper_pptp.c +1 -1
 	DEBUGP("altering call id from 0x%04x to 0x%04x\n",
-		ntohs(*cid), ntohs(new_callid));
+		ntohs(*(u_int16_t *)pptpReq + cid_off), ntohs(new_callid));

Note that this fix introduces another bug in the above use DEBUGP statement, as there is (u_int16_t *) ptr arithmetic used, whereas cid_off is a byte offset.

A fix for that was send a few weeks ago on netfilter-devel.

Great, care to forward it to [email protected] so we can get it in the next release?

Quoted 5 lines
Quoted 3 lines

I have a copy of the patch and I'll cc stable@ on it. Although afaik this bug just causes wrong debug info to come out, so I don't think it's terribly important (?)

If that's the only problem with it, no it's not worth adding to -stable.

Yes, this patch only fixes code in DEBUG statements. Debug can only be enabled at compile time, so I agree it's not a candidate for -stable.

Signature · 5 lines
- Harald Welte <[email protected]>                 http://netfilter.org/
============================================================================
  "Fragmentation is like classful addressing -- an interesting early
   architectural error that shows how much experimentation was going
   on while IP was being designed."                    -- Paul Vixie
#12 lore raw
In reply to Greg KH
Quoted 5 lines · Greg KH

This is the start of the stable review cycle for the 2.6.14.7 release. There are 6 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let us know. If anyone is a maintainer of the proper subsystem, and wants to add a signed-off-by: line to the patch, please respond with it.

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#13 lore raw
Quoted 4 lines · Chuck Wolber

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

I don't see a problems with doing additional stable releases for any kernel, I just wouldn't commit to supporting any specific number of releases. Basically if people send enough patches to warrant a review/release there is obviously some interest. What is the harm?

Justin

#14 lore raw
Quoted 9 lines · Justin M. Forbes
Quoted 3 lines · Chuck Wolber

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

I don't see a problems with doing additional stable releases for any kernel, I just wouldn't commit to supporting any specific number of releases. Basically if people send enough patches to warrant a review/release there is obviously some interest. What is the harm?

The harm is that stable release patches will eventually start being maintained and we'll have to add another stable release "dot" to the end of the growing width of the release version moniker. This stable branch was meant only for "one-off" fixes to a stable release, not for adding fixes upon fixes upon fixes that eventually turn into features that have to be maintained. A new stable release means we change our focus to it and ignore the old stable release.

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#15 lore raw
Quoted 12 lines · On Sat, 28 Jan 2006 20:52:46 -0800 (PST) Chuck Wolber wrote:
Quoted 9 lines · Justin M. Forbes
Quoted 3 lines · Chuck Wolber

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

I don't see a problems with doing additional stable releases for any kernel, I just wouldn't commit to supporting any specific number of releases. Basically if people send enough patches to warrant a review/release there is obviously some interest. What is the harm?

Quoted 7 lines

The harm is that stable release patches will eventually start being maintained and we'll have to add another stable release "dot" to the end of the growing width of the release version moniker. This stable branch was meant only for "one-off" fixes to a stable release, not for adding fixes upon fixes upon fixes that eventually turn into features that have to be maintained. A new stable release means we change our focus to it and ignore the old stable release.

It's a 6-month sliding window for stable releases IIRC. Maybe <[email protected]> can add something like that to Documentation/stable_kernel_rules.txt>.

---
~Randy

author
#16 lore raw
Quoted 22 lines · Randy.Dunlap
Quoted 12 lines · On Sat, 28 Jan 2006 20:52:46 -0800 (PST) Chuck Wolber wrote:
Quoted 9 lines · Justin M. Forbes
Quoted 3 lines · Chuck Wolber

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

I don't see a problems with doing additional stable releases for any kernel, I just wouldn't commit to supporting any specific number of releases. Basically if people send enough patches to warrant a review/release there is obviously some interest. What is the harm?

Quoted 7 lines

The harm is that stable release patches will eventually start being maintained and we'll have to add another stable release "dot" to the end of the growing width of the release version moniker. This stable branch was meant only for "one-off" fixes to a stable release, not for adding fixes upon fixes upon fixes that eventually turn into features that have to be maintained. A new stable release means we change our focus to it and ignore the old stable release.

Quoted 3 lines

It's a 6-month sliding window for stable releases IIRC. Maybe <[email protected]> can add something like that to Documentation/stable_kernel_rules.txt>.

No, it's not a 6 month window, I released this because people sent us patches that they said should go into the 2.6.14-stable tree. And as people complained so much on lkml that we were dropping the old kernels too fast, I never thought that people would complain that we are maintaining older stuff that people seem interested in...

And, odds are, it will probably be the last 2.6.14 stable kernel we (the stable team) release, unless something unusual happens...

And, as always, anyone is free to take on maintaining any of the different kernel versions for as long as they wish.

Does that help?

Man, people complain when you don't maintain older kernels, and they complain when you do...

thanks,

greg k-h

#17 lore raw
Quoted 36 lines · Greg KH
Quoted 22 lines · Randy.Dunlap
Quoted 12 lines · On Sat, 28 Jan 2006 20:52:46 -0800 (PST) Chuck Wolber wrote:
Quoted 9 lines · Justin M. Forbes
Quoted 3 lines · Chuck Wolber

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

I don't see a problems with doing additional stable releases for any kernel, I just wouldn't commit to supporting any specific number of releases. Basically if people send enough patches to warrant a review/release there is obviously some interest. What is the harm?

Quoted 7 lines

The harm is that stable release patches will eventually start being maintained and we'll have to add another stable release "dot" to the end of the growing width of the release version moniker. This stable branch was meant only for "one-off" fixes to a stable release, not for adding fixes upon fixes upon fixes that eventually turn into features that have to be maintained. A new stable release means we change our focus to it and ignore the old stable release.

Quoted 3 lines

It's a 6-month sliding window for stable releases IIRC. Maybe <[email protected]> can add something like that to Documentation/stable_kernel_rules.txt>.

No, it's not a 6 month window, I released this because people sent us patches that they said should go into the 2.6.14-stable tree. And as people complained so much on lkml that we were dropping the old kernels too fast, I never thought that people would complain that we are maintaining older stuff that people seem interested in...

And, odds are, it will probably be the last 2.6.14 stable kernel we (the stable team) release, unless something unusual happens...

Quoted 7 lines

And, as always, anyone is free to take on maintaining any of the different kernel versions for as long as they wish.

Does that help?

Man, people complain when you don't maintain older kernels, and they complain when you do...

Greg, there will always be stupid people who don't understand the work of others. These are the same type of people who won't understand at all why there's a -stable branch. When I started 2.4-hf, I've been told "you're dumb, 2.4 is dead". I'm very glad that you take care of people who cannot easily upgrade to latest version, and I'm sure that a lot of users are too.

Quoted 3 lines

thanks,

greg k-h

Thanks for keeping up the good work,
Willy

#18 lore raw
Quoted 2 lines · Willy Tarreau

Greg, there will always be stupid people who don't understand the work of others.

Hmmm, true, but I wouldn't call them stupid, just a bit ignorant. It's incumbent upon us to manage and maintain expectations to fix that problem.

Quoted 2 lines

I'm very glad that you take care of people who cannot easily upgrade to latest version, and I'm sure that a lot of users are too.

Me too. I'd never dream of criticizing Greg's efforts. He's done an incredible job. I think better defining -stable will make life easier for him and help us be more productive.

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#19 lore raw
In reply to Greg KH
Quoted 5 lines · Greg KH

No, it's not a 6 month window, I released this because people sent us patches that they said should go into the 2.6.14-stable tree. And as people complained so much on lkml that we were dropping the old kernels too fast, I never thought that people would complain that we are maintaining older stuff that people seem interested in...

Don't think of it as a complaint. I'm sorry if it came off that way. It's anyone's job if they want to maintain any kernel release as long as they see fit.

I guess it's how we want to define ourselves. Was I mistaken to think that the -stable team maintains -stable for the current patch release only?

Quoted 4 lines

And, as always, anyone is free to take on maintaining any of the different kernel versions for as long as they wish.

Does that help?

Well, that's the way it's always been. We have a late model 2.5 kernel that we still maintain in a level C avionics device on Boeing jets. You'd be amazed at how stable a 2.5 kernel can be made ;)

Quoted 2 lines

Man, people complain when you don't maintain older kernels, and they complain when you do...

Nope, wasn't complaining. It was an attempt at a dialog to more clearly define our purpose (or perhaps my personal understanding of it) *AND* let other developers know what to expect of us. We can choose to take on as much or as little as we want. I personally am good with whatever everyone else is comfortable taking on. We should manage outside expectations though so you don't have to answer "why do you drop old kernels" questions on your own.

Things have changed a lot with the 2.6 release series and -stable seems to me to be one of those growing pains that still isn't fully resolved. It's a blending between the traditional stable release maintainers and the more nimble release schedule these days. Can the -stable team fill the void, or do we limit our scope to certain patch releases?

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#20 lore raw
In reply to Justin M. Forbes
Quoted 9 lines · Justin M. Forbes
Quoted 3 lines · Chuck Wolber

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

I don't see a problems with doing additional stable releases for any kernel, I just wouldn't commit to supporting any specific number of releases. Basically if people send enough patches to warrant a review/release there is obviously some interest. What is the harm?

No harm at all, and that settles it for me then. I like the idea of having a demand based release system. If no patches arrive, no releases go out.

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#21 lore raw
In reply to Chuck Wolber
Quoted 10 lines · On Sat, 28 Jan 2006 20:30:25 -0800 (PST) Chuck Wolber wrote:
Quoted 5 lines · Greg KH

This is the start of the stable review cycle for the 2.6.14.7 release. There are 6 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let us know. If anyone is a maintainer of the proper subsystem, and wants to add a signed-off-by: line to the patch, please respond with it.

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

Some people wanted more -stable so the stable team agreed to do a little more. Is it a problem?

---
~Randy

#22 lore raw
Quoted 11 lines · Randy.Dunlap
Quoted 9 lines · On Sat, 28 Jan 2006 20:30:25 -0800 (PST) Chuck Wolber wrote:
Quoted 6 lines · Greg KH

This is the start of the stable review cycle for the 2.6.14.7 release. There are 6 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let us know. If anyone is a maintainer of the proper subsystem, and wants to add a signed-off-by: line to the patch, please respond with it.

Quoted 5 lines
Quoted 2 lines

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

Some people wanted more -stable so the stable team agreed to do a little more. Is it a problem?

I don't know if there is a problem, but it goes against the concept of "one-off" fixes that aren't maintained (aka the purpose of the -stable team). This slope eventually leads us to backporting -stable fixes from other -stable releases etc etc.

If there's one thing I've learned from watching guys like Alan Cox maintain stable releases, it's that they're profoundly good at saying "no". I'm not saying that's warranted here, I'm just trying to encourage the dialog (I suspect that I've missed part of the conversation as I am not currently subscribed to LKML).

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#23 lore raw
Quoted 16 lines · Chuck Wolber
Quoted 14 lines · Randy.Dunlap
Quoted 11 lines · On Sat, 28 Jan 2006 20:30:25 -0800 (PST) Chuck Wolber wrote:
Quoted 6 lines · Greg KH

This is the start of the stable review cycle for the 2.6.14.7 release. There are 6 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let us know. If anyone is a maintainer of the proper subsystem, and wants to add a signed-off-by: line to the patch, please respond with it.

Please correct me if I'm wrong here, but aren't we supposed to stop doing this for the 2.6.14 release now that 2.6.15 is out?

Quoted 8 lines
Quoted 2 lines

Some people wanted more -stable so the stable team agreed to do a little more. Is it a problem?

I don't know if there is a problem, but it goes against the concept of "one-off" fixes that aren't maintained (aka the purpose of the -stable team). This slope eventually leads us to backporting -stable fixes from other -stable releases etc etc.

The purpose of -stable is to provide stable kernels to 2.6 users. If time was not a problem, it's possible that there would be even more versions supported. The day you will install Linux on a server, you'll understand why it's problematic for some people to upgrade to latest version to get fixes. And when you get something that works, you hope to be able to use it as an alternative to a simple upgrade when the later breaks on your hardware.

Quoted 5 lines

If there's one thing I've learned from watching guys like Alan Cox maintain stable releases, it's that they're profoundly good at saying "no". I'm not saying that's warranted here, I'm just trying to encourage the dialog (I suspect that I've missed part of the conversation as I am not currently subscribed to LKML).

It's not a matter of saying "yes" or "no", it's a matter of helping users in getting something which works best on their hardware while still being reliable and secure. Maintainers propose some solutions for this, and can adapt to users' demands. I don't see anything wrong with this.

Quoted 1 line

..Chuck..

Willy

#24 lore raw
Quoted 3 lines · Willy Tarreau

The purpose of -stable is to provide stable kernels to 2.6 users. If time was not a problem, it's possible that there would be even more versions supported.

Mmmm, yup. I was there when this whole thing was brewed up ;)

Quoted 2 lines

The day you will install Linux on a server, you'll understand why it's problematic for some people to upgrade to latest version to get fixes.

Sure hope you're not implying I haven't done that before...

Quoted 5 lines

It's not a matter of saying "yes" or "no", it's a matter of helping users in getting something which works best on their hardware while still being reliable and secure. Maintainers propose some solutions for this, and can adapt to users' demands. I don't see anything wrong with this.

Neither do I. To a great extent, that void is already filled by distro maintainers.

..Chuck..

Signature · 6 lines
http://www.quantumlinux.com
 Quantum Linux Laboratories, LLC.
 ACCELERATING Business with Open Technology

 "The measure of the restoration lies in the extent to which we apply
  social values more noble than mere monetary profit." - FDR
#25 lore raw
In reply to Chuck Wolber
Quoted 2 lines · Chuck Wolber

I don't know if there is a problem, but it goes against the concept of "one-off" fixes that aren't maintained

If they have time for this then great, I can't see any drawbacks. The early concept was just meant to limit the workload.

Quoted 3 lines

(aka the purpose of the -stable
team). This slope eventually leads us to backporting -stable fixes from other -stable releases etc etc.

As long as they can contribute their time it's a win.

Signature · 1 line
Krzysztof Halasa