3.0.32-stable review

This is the start of the stable review cycle for the 3.0.32 release. There are 54 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 me know.

Responses should be made by Sun May 20 21:15:57 UTC 2012. Anything received after that time might be too late.

The whole patch series can be found in one patch at:

	kernel.org/pub/linux/kernel/v3.0/stable-review/patch-3.0.32-rc1.gz

and the diffstat can be found below.

thanks,

greg k-h

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

 Documentation/networking/ip-sysctl.txt |    4 +-
 Makefile                               |    4 +-
 arch/arm/kernel/smp.c                  |    4 +-
 arch/arm/kernel/sys_arm.c              |    2 +-
 arch/arm/mach-orion5x/mpp.h            |    4 +-
 arch/arm/mm/fault.c                    |    4 +-
 arch/sparc/kernel/central.c            |    2 +-
 arch/sparc/mm/ultra.S                  |    6 +--
 arch/tile/kernel/compat_signal.c       |   12 ++---
 arch/x86/kernel/setup_percpu.c         |   14 +++++-
 arch/x86/xen/enlighten.c               |    7 ++-
 arch/x86/xen/mmu.c                     |    7 ++-
 drivers/acpi/sleep.c                   |    8 ++++
 drivers/crypto/Kconfig                 |    1 +
 drivers/gpio/Kconfig                   |   11 +++--
 drivers/gpio/pch_gpio.c                |    1 +
 drivers/md/md.c                        |    2 +
 drivers/media/rc/ene_ir.c              |   32 ++++++-------
 drivers/media/rc/fintek-cir.c          |   20 ++++----
 drivers/media/rc/ite-cir.c             |   20 ++++----
 drivers/media/rc/nuvoton-cir.c         |   36 +++++++--------
 drivers/media/rc/winbond-cir.c         |   78 ++++++++++++++++----------------
 drivers/net/Kconfig                    |   11 +++--
 drivers/net/pch_gbe/pch_gbe_main.c     |   50 +++++++++++++++-----
 drivers/net/pch_gbe/pch_gbe_param.c    |   15 +++---
 drivers/net/sky2.c                     |   31 ++++++++-----
 drivers/net/sky2.h                     |    1 -
 drivers/net/sungem.c                   |    2 +-
 drivers/net/tg3.c                      |   18 +++++++-
 drivers/net/usb/asix.c                 |    4 +-
 drivers/net/usb/cdc_ether.c            |   30 +++++++++++-
 drivers/net/usb/smsc95xx.c             |    2 +-
 drivers/net/usb/usbnet.c               |   54 +++++++++++++++-------
 drivers/platform/x86/sony-laptop.c     |    2 +-
 drivers/regulator/max8997.c            |    2 +-
 drivers/scsi/hpsa.c                    |    4 +-
 fs/ext3/ialloc.c                       |    8 +++-
 fs/ext4/ialloc.c                       |    8 +++-
 fs/ext4/xattr.c                        |    7 +--
 fs/jffs2/gc.c                          |    2 +-
 fs/nfs/nfs4proc.c                      |    1 +
 fs/super.c                             |    2 +
 include/asm-generic/statfs.h           |    2 +-
 include/linux/seqlock.h                |    2 +-
 include/linux/usb/usbnet.h             |    3 +-
 init/do_mounts.c                       |    2 +-
 kernel/compat.c                        |   63 +++++++++++++++++++-------
 kernel/fork.c                          |    3 ++
 mm/hugetlb.c                           |    1 -
 mm/memcontrol.c                        |    6 +++
 mm/nobootmem.c                         |    3 +-
 mm/percpu.c                            |   12 +++++
 net/core/dev.c                         |   20 ++++++++
 net/ipv4/tcp.c                         |   12 ++---
 net/ipv4/tcp_input.c                   |    2 +-
 net/l2tp/l2tp_ip.c                     |    3 +-
 net/sched/sch_netem.c                  |    6 +--
 sound/pci/echoaudio/echoaudio_dsp.c    |    2 +-
 sound/pci/hda/hda_intel.c              |    6 ++-
 59 files changed, 448 insertions(+), 233 deletions(-)

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

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

From: Paolo Pisati <[email protected]>

commit 07d69d4238418746a7b85c5d05ec17c658a2a390 upstream.

Without this patch sysfs reports the cable as present

flag@flag-desktop:~$ cat /sys/class/net/eth0/carrier
1

while it's not:

flag@flag-desktop:~$ sudo mii-tool eth0
eth0: no link

Tested on my Beagle XM.

v2: added mantainer to the list of recipient

---
 drivers/net/usb/smsc95xx.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -1190,7 +1190,7 @@ static const struct driver_info smsc95xx
 	.rx_fixup	= smsc95xx_rx_fixup,
 	.tx_fixup	= smsc95xx_tx_fixup,
 	.status		= smsc95xx_status,
-	.flags		= FLAG_ETHER | FLAG_SEND_ZLP,
+	.flags		= FLAG_ETHER | FLAG_SEND_ZLP | FLAG_LINK_INTR,
 };
 
 static const struct usb_device_id products[] = {

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

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

From: Konrad Rzeszutek Wilk <[email protected]>

commit b7e5ffe5d83fa40d702976d77452004abbe35791 upstream.

If I try to do "cat /sys/kernel/debug/kernel_page_tables" I end up with:

BUG: unable to handle kernel paging request at ffffc7fffffff000
IP: [<ffffffff8106aa51>] ptdump_show+0x221/0x480
PGD 0
Oops: 0000 [#1] SMP

CPU 0
.. snip..
RAX: 0000000000000000 RBX: ffffc00000000fff RCX: 0000000000000000
RDX: 0000800000000000 RSI: 0000000000000000 RDI: ffffc7fffffff000

which is due to the fact we are trying to access a PFN that is not accessible to us. The reason (at least in this case) was that PGD[256] is set to __HYPERVISOR_VIRT_START which was setup (by the hypervisor) to point to a read-only linear map of the MFN->PFN array. During our parsing we would get the MFN (a valid one), try to look it up in the MFN->PFN tree and find it invalid and return ~0 as PFN. Then pte_mfn_to_pfn would happilly feed that in, attach the flags and return it back to the caller. 'ptdump_show' bitshifts it and gets and invalid value that it tries to dereference.

Instead of doing all of that, we detect the ~0 case and just return !_PAGE_PRESENT.

This bug has been in existence .. at least until 2.6.37 (yikes!)

---
 arch/x86/xen/mmu.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

@@ -320,8 +320,13 @@ static pteval_t pte_mfn_to_pfn(pteval_t
 {
 	if (val & _PAGE_PRESENT) {
 		unsigned long mfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
+		unsigned long pfn = mfn_to_pfn(mfn);
+
 		pteval_t flags = val & PTE_FLAGS_MASK;
-		val = ((pteval_t)mfn_to_pfn(mfn) << PAGE_SHIFT) | flags;
+		if (unlikely(pfn == ~0))
+			val = flags & ~_PAGE_PRESENT;
+		else
+			val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
 	}
 
 	return val;

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

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

From: David Vrabel <[email protected]>

commit 76a8df7b49168509df02461f83fab117a4a86e08 upstream.

The accessing PCI configuration space with the PCI BIOS32 service does not work in PV guests.

On systems without MMCONFIG or where the BIOS hasn't marked the MMCONFIG region as reserved in the e820 map, the BIOS service is probed (even though direct access is preferred) and this hangs.

---
 arch/x86/xen/enlighten.c |    7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

@@ -62,6 +62,7 @@
 #include <asm/reboot.h>
 #include <asm/stackprotector.h>
 #include <asm/hypervisor.h>
+#include <asm/pci_x86.h>
 
 #include "xen-ops.h"
 #include "mmu.h"

@@ -1259,8 +1260,10 @@ asmlinkage void __init xen_start_kernel(
 		/* Make sure ACS will be enabled */
 		pci_request_acs();
 	}
-		
-
+#ifdef CONFIG_PCI
+	/* PCI BIOS service won't work from a PV guest. */
+	pci_probe &= ~PCI_PROBE_BIOS;
+#endif
 	xen_raw_console_write("about to get started...\n");
 
 	xen_setup_runstate_info(0);

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

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

From: Tejun Heo <[email protected]>

commit d5e28005a1d2e67833852f4c9ea8ec206ea3ff85 upstream.

With the embed percpu first chunk allocator, x86 uses either PAGE_SIZE or PMD_SIZE for atom_size. PMD_SIZE is used when CPU supports PSE so that percpu areas are aligned to PMD mappings and possibly allow using PMD mappings in vmalloc areas in the future. Using larger atom_size doesn't waste actual memory; however, it does require larger vmalloc space allocation later on for !first chunks.

With reasonably sized vmalloc area, PMD_SIZE shouldn't be a problem but x86_32 at this point is anything but reasonable in terms of address space and using larger atom_size reportedly leads to frequent percpu allocation failures on certain setups.

As there is no reason to not use PMD_SIZE on x86_64 as vmalloc space is aplenty and most x86_64 configurations support PSE, fix the issue by always using PMD_SIZE on x86_64 and PAGE_SIZE on x86_32.

v2: drop cpu_has_pse test and make x86_64 always use PMD_SIZE and
x86_32 PAGE_SIZE as suggested by hpa.

LKML-Reference: <[email protected]>

---
 arch/x86/kernel/setup_percpu.c |   14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

@@ -185,10 +185,22 @@ void __init setup_per_cpu_areas(void)
 #endif
 	rc = -EINVAL;
 	if (pcpu_chosen_fc != PCPU_FC_PAGE) {
-		const size_t atom_size = cpu_has_pse ? PMD_SIZE : PAGE_SIZE;
 		const size_t dyn_size = PERCPU_MODULE_RESERVE +
 			PERCPU_DYNAMIC_RESERVE - PERCPU_FIRST_CHUNK_RESERVE;
+		size_t atom_size;
 
+		/*
+		 * On 64bit, use PMD_SIZE for atom_size so that embedded
+		 * percpu areas are aligned to PMD.  This, in the future,
+		 * can also allow using PMD mappings in vmalloc area.  Use
+		 * PAGE_SIZE on 32bit as vmalloc space is highly contended
+		 * and large vmalloc area allocs can easily fail.
+		 */
+#ifdef CONFIG_X86_64
+		atom_size = PMD_SIZE;
+#else
+		atom_size = PAGE_SIZE;
+#endif
 		rc = pcpu_embed_first_chunk(PERCPU_FIRST_CHUNK_RESERVE,
 					    dyn_size, atom_size,
 					    pcpu_cpu_distance,

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

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

From: "H. Peter Anvin" <[email protected]>

commit f5c2347ee20a8d6964d6a6b1ad04f200f8d4dfa7 upstream.

<asm-generic/statfs.h> is exported to userspace, so using BITS_PER_LONG is invalid. We need to use __BITS_PER_LONG instead.

This is kernel bugzilla 43165.

---
 include/asm-generic/statfs.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -15,7 +15,7 @@ typedef __kernel_fsid_t	fsid_t;
  * with a 10' pole.
  */
 #ifndef __statfs_word
-#if BITS_PER_LONG == 64
+#if __BITS_PER_LONG == 64
 #define __statfs_word long
 #else
 #define __statfs_word __u32

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

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

From: Linus Torvalds <[email protected]>

commit 2f624278626677bfaf73fef97f86b37981621f5c upstream.

We really need to use a ACCESS_ONCE() on the sequence value read in __read_seqcount_begin(), because otherwise the compiler might end up reloading the value in between the test and the return of it. As a result, it might end up returning an odd value (which means that a write is in progress).

If the reader is then fast enough that that odd value is still the current one when the read_seqcount_retry() is done, we might end up with a "successful" read sequence, even despite the concurrent write being active.

In practice this probably never really happens - there just isn't anything else going on around the read of the sequence count, and the common case is that we end up having a read barrier immediately afterwards.

So the code sequence in which gcc might decide to reaload from memory is small, and there's no reason to believe it would ever actually do the reload. But if the compiler ever were to decide to do so, it would be incredibly annoying to debug. Let's just make sure.

---
 include/linux/seqlock.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -141,7 +141,7 @@ static inline unsigned __read_seqcount_b
 	unsigned ret;
 
 repeat:
-	ret = s->sequence;
+	ret = ACCESS_ONCE(s->sequence);
 	if (unlikely(ret & 1)) {
 		cpu_relax();
 		goto repeat;

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

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

From: Tim Bird <[email protected]>

commit e787ec1376e862fcea1bfd523feb7c5fb43ecdb9 upstream.

The inline assembly in kernel_execve() uses r8 and r9. Since this code sequence does not return, it usually doesn't matter if the register clobber list is accurate. However, I saw a case where a particular version of gcc used r8 as an intermediate for the value eventually passed to r9. Because r8 is used in the inline assembly, and not mentioned in the clobber list, r9 was set to an incorrect value.

This resulted in a kernel panic on execution of the first user-space program in the system. r9 is used in ret_to_user as the thread_info pointer, and if it's wrong, bad things happen.

---
 arch/arm/kernel/sys_arm.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -115,7 +115,7 @@ int kernel_execve(const char *filename,
 		  "Ir" (THREAD_START_SP - sizeof(regs)),
 		  "r" (&regs),
 		  "Ir" (sizeof(regs))
-		: "r0", "r1", "r2", "r3", "ip", "lr", "memory");
+		: "r0", "r1", "r2", "r3", "r8", "r9", "ip", "lr", "memory");
 
  out:
 	return ret;

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

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

From: Colin Cross <[email protected]>

commit fde165b2a29673aabf18ceff14dea1f1cfb0daad upstream.

Commit 4e8ee7de227e3ab9a72040b448ad728c5428a042 (ARM: SMP: use idmap_pgd for mapping MMU enable during secondary booting) switched secondary boot to use idmap_pgd, which is initialized during early_initcall, instead of a page table initialized during __cpu_up. This causes idmap_pgd to contain the static mappings but be missing all dynamic mappings.

If a console is registered that creates a dynamic mapping, the printk in secondary_start_kernel will trigger a data abort on the missing mapping before the exception handlers have been initialized, leading to a hang. Initial boot is not affected because no consoles have been registered, and resume is usually not affected because the offending console is suspended. Onlining a cpu with hotplug triggers the problem.

A workaround is to the printk in secondary_start_kernel until after the page tables have been switched back to init_mm.

---
 arch/arm/kernel/smp.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

@@ -279,8 +279,6 @@ asmlinkage void __cpuinit secondary_star
 	struct mm_struct *mm = &init_mm;
 	unsigned int cpu = smp_processor_id();
 
-	printk("CPU%u: Booted secondary processor\n", cpu);
-
 	/*
 	 * All kernel threads share the same mm context; grab a
 	 * reference and switch to it.

@@ -292,6 +290,8 @@ asmlinkage void __cpuinit secondary_star
 	enter_lazy_tlb(mm, current);
 	local_flush_tlb_all();
 
+	printk("CPU%u: Booted secondary processor\n", cpu);
+
 	cpu_init();
 	preempt_disable();
 	trace_hardirqs_off();

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

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

From: Axel Lin <[email protected]>

commit f55205f4d4a8823a11bb8b37ef2ecbd78fb09463 upstream.

I think this is a typo.
To ensure new voltage setting won't greater than desc->max, the equation should be desc->min + desc->step * new_val <= desc->max.

---
 drivers/regulator/max8997.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -688,7 +688,7 @@ static int max8997_set_voltage_buck(stru
 		}
 
 		new_val++;
-	} while (desc->min + desc->step + new_val <= desc->max);
+	} while (desc->min + desc->step * new_val <= desc->max);
 
 	new_idx = tmp_idx;
 	new_val = tmp_val;

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

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

From: Ben Hutchings <[email protected]>

commit 48d99f47a81a66bdd61a348c7fe8df5a7afdf5f3 upstream.

Commit 554cdaefd1cf7bb54b209c4e68c7cec87ce442a9 ('ARM: orion5x: Refactor mpp code to use common orion platform mpp.') seems to have accidentally inverted the GPIO valid bits for MPP9 (only). For the mv2120 platform which uses MPP9 as a GPIO LED device, this results in the error:

[   12.711476] leds-gpio: probe of leds-gpio failed with error -22

References: http://bugs.debian.org/667446

---
 arch/arm/mach-orion5x/mpp.h |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

@@ -65,8 +65,8 @@
 #define MPP8_GIGE               MPP(8,  0x1, 0, 0, 1,   1,   1)
 
 #define MPP9_UNUSED		MPP(9,  0x0, 0, 0, 1,   1,   1)
-#define MPP9_GPIO		MPP(9,  0x0, 0, 0, 1,   1,   1)
-#define MPP9_GIGE               MPP(9,  0x1, 1, 1, 1,   1,   1)
+#define MPP9_GPIO		MPP(9,  0x0, 1, 1, 1,   1,   1)
+#define MPP9_GIGE               MPP(9,  0x1, 0, 0, 1,   1,   1)
 
 #define MPP10_UNUSED		MPP(10, 0x0, 0, 0, 1,   1,   1)
 #define MPP10_GPIO		MPP(10, 0x0, 1, 1, 1,   1,   1)

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

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

From: Ingo van Lil <[email protected]>

[ Upstream commit 2a5809499e35b53a6044fd34e72b242688b7a862 ]

The asix.c USB Ethernet driver avoids ending a tx transfer with a zero- length packet by appending a four-byte padding to transfers whose length is a multiple of maxpacket. However, the hard-coded 512 byte maxpacket length is valid for high-speed USB only; full-speed USB uses 64 byte packets.

---
 drivers/net/usb/asix.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

@@ -398,7 +398,7 @@ static struct sk_buff *asix_tx_fixup(str
 	u32 packet_len;
 	u32 padbytes = 0xffff0000;
 
-	padlen = ((skb->len + 4) % 512) ? 0 : 4;
+	padlen = ((skb->len + 4) & (dev->maxpacket - 1)) ? 0 : 4;
 
 	if ((!skb_cloned(skb)) &&
 	    ((headroom + tailroom) >= (4 + padlen))) {

@@ -420,7 +420,7 @@ static struct sk_buff *asix_tx_fixup(str
 	cpu_to_le32s(&packet_len);
 	skb_copy_to_linear_data(skb, &packet_len, sizeof(packet_len));
 
-	if ((skb->len % 512) == 0) {
+	if (padlen) {
 		cpu_to_le32s(&padbytes);
 		memcpy(skb_tail_pointer(skb), &padbytes, sizeof(padbytes));
 		skb_put(skb, sizeof(padbytes));

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

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

From: Eric Dumazet <[email protected]>

[ Upstream commit 116a0fc31c6c9b8fc821be5a96e5bf0b43260131 ]

skb_checksum_help(skb) can return an error, we must free skb in this case. qdisc_drop(skb, sch) can also be feeded with a NULL skb (if skb_unshare() failed), so lets use this generic helper.

---
 net/sched/sch_netem.c |    6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

@@ -350,10 +350,8 @@ static int netem_enqueue(struct sk_buff
 	if (q->corrupt && q->corrupt >= get_crandom(&q->corrupt_cor)) {
 		if (!(skb = skb_unshare(skb, GFP_ATOMIC)) ||
 		    (skb->ip_summed == CHECKSUM_PARTIAL &&
-		     skb_checksum_help(skb))) {
-			sch->qstats.drops++;
-			return NET_XMIT_DROP;
-		}
+		     skb_checksum_help(skb)))
+			return qdisc_drop(skb, sch);
 
 		skb->data[net_random() % skb_headlen(skb)] ^= 1<<(net_random() % 8);
 	}

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

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

From: "Eric W. Biederman" <[email protected]>

[ Upstream commit 7d3d43dab4e978d8d9ad1acf8af15c9b1c4b0f0f ]

We already synthesize events in register_netdevice_notifier and synthesizing events in unregister_netdevice_notifier allows to us remove the need for special case cleanup code.

This change should be safe as it adds no new cases for existing callers of unregiser_netdevice_notifier to handle.

---
 net/core/dev.c |   20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

@@ -1406,14 +1406,34 @@ EXPORT_SYMBOL(register_netdevice_notifie
  *	register_netdevice_notifier(). The notifier is unlinked into the
  *	kernel structures and may then be reused. A negative errno code
  *	is returned on a failure.
+ *
+ * 	After unregistering unregister and down device events are synthesized
+ *	for all devices on the device list to the removed notifier to remove
+ *	the need for special case cleanup code.
  */
 
 int unregister_netdevice_notifier(struct notifier_block *nb)
 {
+	struct net_device *dev;
+	struct net *net;
 	int err;
 
 	rtnl_lock();
 	err = raw_notifier_chain_unregister(&netdev_chain, nb);
+	if (err)
+		goto unlock;
+
+	for_each_net(net) {
+		for_each_netdev(net, dev) {
+			if (dev->flags & IFF_UP) {
+				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
+				nb->notifier_call(nb, NETDEV_DOWN, dev);
+			}
+			nb->notifier_call(nb, NETDEV_UNREGISTER, dev);
+			nb->notifier_call(nb, NETDEV_UNREGISTER_BATCH, dev);
+		}
+	}
+unlock:
 	rtnl_unlock();
 	return err;
 }

Please also include these two commits in addition to this change, to avoid a regression with pktgen:
commit c57b54684060c8aced64a5b78ff69ff289af97b9
commit d4b1133558e0d417342d5d2c49e4c35b428ff20d

[]'s
Herton

I need an ack from the network maintainers for me to be able to accept this.

thanks,

greg k-h

From: Greg KH <[email protected]>
Date: Sun, 27 May 2012 09:13:49 +0900

These bugs only effect networking hackers doing low-level packet spamming on their network using pktgen, who furthermore try to unload the pktgen module.

Low level network packet spamming using pktgen, and trying to unload that module, adds up to an extremely small minority of users.

Therefore these changes are not worth putting into -stable by any measurement.

Ben decided to add them to 3.2.x, but that is entirely his perogative.

Ok, thanks, I'll not do this for 3.0-stable then, as no developer should really be using that kernel anymore :)

greg k-h

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

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

From: Sasha Levin <[email protected]>

[ Upstream commit 84768edbb2721637620b2d84501bb0d5aed603f1 ]

l2tp_ip_sendmsg could return without releasing socket lock, making it all the way to userspace, and generating the following warning:

[  130.891594] ================================================
[  130.894569] [ BUG: lock held when returning to user space! ]
[  130.897257] 3.4.0-rc5-next-20120501-sasha #104 Tainted: G        W
[  130.900336] ------------------------------------------------
[  130.902996] trinity/8384 is leaving the kernel with locks still held!
[  130.906106] 1 lock held by trinity/8384:
[  130.907924]  #0:  (sk_lock-AF_INET){+.+.+.}, at: [<ffffffff82b9503f>] l2tp_ip_sendmsg+0x2f/0x550

Introduced by commit 2f16270 ("l2tp: Fix locking in l2tp_ip.c").

---
 net/l2tp/l2tp_ip.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

@@ -441,8 +441,9 @@ static int l2tp_ip_sendmsg(struct kiocb
 
 		daddr = lip->l2tp_addr.s_addr;
 	} else {
+		rc = -EDESTADDRREQ;
 		if (sk->sk_state != TCP_ESTABLISHED)
-			return -EDESTADDRREQ;
+			goto out;
 
 		daddr = inet->inet_daddr;
 		connected = 1;

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

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

From: stephen hemminger <[email protected]>

[ Upstream commit 3f42941b5d1d13542b1a755a9e4f633aa72e4d3e ]

When a small packet is received, the driver copies it to a new skb to allow reusing the full size Rx buffer. The copy was propogating the checksum offload but not the receive hash information. The bug is impact was mostly harmless and therefore not observed until reviewing this area of code.

---
 drivers/net/sky2.c |    3 +++
 1 file changed, 3 insertions(+)

@@ -2344,8 +2344,11 @@ static struct sk_buff *receive_copy(stru
 		skb_copy_from_linear_data(re->skb, skb->data, length);
 		skb->ip_summed = re->skb->ip_summed;
 		skb->csum = re->skb->csum;
+		skb->rxhash = re->skb->rxhash;
+
 		pci_dma_sync_single_for_device(sky2->hw->pdev, re->data_addr,
 					       length, PCI_DMA_FROMDEVICE);
+		re->skb->rxhash = 0;
 		re->skb->ip_summed = CHECKSUM_NONE;
 		skb_put(skb, length);
 	}

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

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

From: stephen hemminger <[email protected]>

[ Upstream commit e072b3fad5f3915102c94628b4971f52ff99dd05 ]

Bug: The VLAN bit of the MAC RX Status Word is unreliable in several older supported chips. Sometimes the VLAN bit is not set for valid VLAN packets and also sometimes the VLAN bit is set for non-VLAN packets that came after a VLAN packet. This results in a receive length error when VLAN hardware tagging is enabled.

Fix: Variation on original fix proposed by Mirko.
The VLAN information is decoded in the status loop, and can be applied to the received SKB there. This eliminates the need for the separate tag field in the interface data structure. The tag has to be copied and cleared if packet is copied. This version checked out with vlan and normal traffic.

Note: vlan_tx_tag_present should be renamed vlan_tag_present, but that is outside scope of this.

---
 drivers/net/sky2.c |   28 +++++++++++++++++-----------
 drivers/net/sky2.h |    1 -
 2 files changed, 17 insertions(+), 12 deletions(-)

@@ -2345,9 +2345,11 @@ static struct sk_buff *receive_copy(stru
 		skb->ip_summed = re->skb->ip_summed;
 		skb->csum = re->skb->csum;
 		skb->rxhash = re->skb->rxhash;
+		skb->vlan_tci = re->skb->vlan_tci;
 
 		pci_dma_sync_single_for_device(sky2->hw->pdev, re->data_addr,
 					       length, PCI_DMA_FROMDEVICE);
+		re->skb->vlan_tci = 0;
 		re->skb->rxhash = 0;
 		re->skb->ip_summed = CHECKSUM_NONE;
 		skb_put(skb, length);

@@ -2433,9 +2435,6 @@ static struct sk_buff *sky2_receive(stru
 	struct sk_buff *skb = NULL;
 	u16 count = (status & GMR_FS_LEN) >> 16;
 
-	if (status & GMR_FS_VLAN)
-		count -= VLAN_HLEN;	/* Account for vlan tag */
-
 	netif_printk(sky2, rx_status, KERN_DEBUG, dev,
 		     "rx slot %u status 0x%x len %d\n",
 		     sky2->rx_next, status, length);

@@ -2443,6 +2442,9 @@ static struct sk_buff *sky2_receive(stru
 	sky2->rx_next = (sky2->rx_next + 1) % sky2->rx_pending;
 	prefetch(sky2->rx_ring + sky2->rx_next);
 
+	if (vlan_tx_tag_present(re->skb))
+		count -= VLAN_HLEN;	/* Account for vlan tag */
+
 	/* This chip has hardware problems that generates bogus status.
 	 * So do only marginal checking and expect higher level protocols
 	 * to handle crap frames.

@@ -2500,11 +2502,8 @@ static inline void sky2_tx_done(struct n
 }
 
 static inline void sky2_skb_rx(const struct sky2_port *sky2,
-			       u32 status, struct sk_buff *skb)
+			       struct sk_buff *skb)
 {
-	if (status & GMR_FS_VLAN)
-		__vlan_hwaccel_put_tag(skb, be16_to_cpu(sky2->rx_tag));
-
 	if (skb->ip_summed == CHECKSUM_NONE)
 		netif_receive_skb(skb);
 	else

@@ -2558,6 +2557,14 @@ static void sky2_rx_checksum(struct sky2
 	}
 }
 
+static void sky2_rx_tag(struct sky2_port *sky2, u16 length)
+{
+	struct sk_buff *skb;
+
+	skb = sky2->rx_ring[sky2->rx_next].skb;
+	__vlan_hwaccel_put_tag(skb, be16_to_cpu(length));
+}
+
 static void sky2_rx_hash(struct sky2_port *sky2, u32 status)
 {
 	struct sk_buff *skb;

@@ -2616,8 +2623,7 @@ static int sky2_status_intr(struct sky2_
 			}
 
 			skb->protocol = eth_type_trans(skb, dev);
-
-			sky2_skb_rx(sky2, status, skb);
+			sky2_skb_rx(sky2, skb);
 
 			/* Stop after net poll weight */
 			if (++work_done >= to_do)

@@ -2625,11 +2631,11 @@ static int sky2_status_intr(struct sky2_
 			break;
 
 		case OP_RXVLAN:
-			sky2->rx_tag = length;
+			sky2_rx_tag(sky2, length);
 			break;
 
 		case OP_RXCHKSVLAN:
-			sky2->rx_tag = length;
+			sky2_rx_tag(sky2, length);
 			/* fall through */
 		case OP_RXCHKS:
 			if (likely(dev->features & NETIF_F_RXCSUM))

@@ -2236,7 +2236,6 @@ struct sky2_port {
 	u16		     rx_pending;
 	u16		     rx_data_size;
 	u16		     rx_nfrags;
-	u16		     rx_tag;
 
 	struct {
 		unsigned long last;

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

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

From: Matt Carlson <[email protected]>

[ Upstream commit f891ea1634ce41f5f47ae40d8594809f4cd2ca66 ]

When RSS is enabled, interrupt vector 0 does not receive any rx traffic. The rx producer index fields for vector 0's status block should be considered reserved in this case. This patch changes the code to respect these reserved fields, which avoids a kernel panic when these fields take on non-zero values.

---
 drivers/net/tg3.c |   18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

@@ -740,8 +740,13 @@ static inline unsigned int tg3_has_work(
 		if (sblk->status & SD_STATUS_LINK_CHG)
 			work_exists = 1;
 	}
-	/* check for RX/TX work to do */
-	if (sblk->idx[0].tx_consumer != tnapi->tx_cons ||
+
+	/* check for TX work to do */
+	if (sblk->idx[0].tx_consumer != tnapi->tx_cons)
+		work_exists = 1;
+
+	/* check for RX work to do */
+	if (tnapi->rx_rcb_prod_idx &&
 	    *(tnapi->rx_rcb_prod_idx) != tnapi->rx_rcb_ptr)
 		work_exists = 1;
 

@@ -5216,6 +5221,9 @@ static int tg3_poll_work(struct tg3_napi
 			return work_done;
 	}
 
+	if (!tnapi->rx_rcb_prod_idx)
+		return work_done;
+
 	/* run RX thread, within the bounds set by NAPI.
 	 * All RX "locking" is done by ensuring outside
 	 * code synchronizes with tg3->napi.poll()

@@ -6626,6 +6634,12 @@ static int tg3_alloc_consistent(struct t
 		 */
 		switch (i) {
 		default:
+			if (tg3_flag(tp, ENABLE_RSS)) {
+				tnapi->rx_rcb_prod_idx = NULL;
+				break;
+			}
+			/* Fall through */
+		case 1:
 			tnapi->rx_rcb_prod_idx = &sblk->idx[0].rx_producer;
 			break;
 		case 2:

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

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

From: Gerard Lledo <[email protected]>

[ Upstream commit 5a8887d39e1ba5ee2d4ccb94b14d6f2dce5ddfca ]

WakeOnLan was broken in this driver because gp->asleep_wol is a 1-bit bitfield and it was being assigned WAKE_MAGIC, which is (1 << 5). gp->asleep_wol remains 0 and the machine never wakes up. Fixed by casting gp->wake_on_lan to bool. Tested on an iBook G4.

---
 drivers/net/sungem.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -2363,7 +2363,7 @@ static int gem_suspend(struct pci_dev *p
 		netif_device_detach(dev);
 
 		/* Switch off MAC, remember WOL setting */
-		gp->asleep_wol = gp->wake_on_lan;
+		gp->asleep_wol = !!gp->wake_on_lan;
 		gem_do_stop(dev, gp->asleep_wol);
 	} else
 		gp->asleep_wol = 0;

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

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

From: Eric Dumazet <[email protected]>

[ Upstream commit b49960a05e32121d29316cfdf653894b88ac9190 ]

tcp_adv_win_scale default value is 2, meaning we expect a good citizen skb to have skb->len / skb->truesize ratio of 75% (3/4)

In 2.6 kernels we (mis)accounted for typical MSS=1460 frame :
1536 + 64 + 256 = 1856 'estimated truesize', and 1856 * 3/4 = 1392. So these skbs were considered as not bloated.

With recent truesize fixes, a typical MSS=1460 frame truesize is now the more precise :
2048 + 256 = 2304. But 2304 * 3/4 = 1728.
So these skb are not good citizen anymore, because 1460 < 1728

(GRO can escape this problem because it build skbs with a too low truesize.)

This also means tcp advertises a too optimistic window for a given allocated rcvspace : When receiving frames, sk_rmem_alloc can hit sk_rcvbuf limit and we call tcp_prune_queue()/tcp_collapse() too often, especially when application is slow to drain its receive queue or in case of losses (netperf is fast, scp is slow). This is a major latency source.

We should adjust the len/truesize ratio to 50% instead of 75%

This patch :

1) changes tcp_adv_win_scale default to 1 instead of 2

2) increase tcp_rmem[2] limit from 4MB to 6MB to take into account better truesize tracking and to allow autotuning tcp receive window to reach same value than before. Note that same amount of kernel memory is consumed compared to 2.6 kernels.

---
 Documentation/networking/ip-sysctl.txt |    4 ++--
 net/ipv4/tcp.c                         |    9 +++++----
 net/ipv4/tcp_input.c                   |    2 +-
 3 files changed, 8 insertions(+), 7 deletions(-)

@@ -147,7 +147,7 @@ tcp_adv_win_scale - INTEGER
 	(if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
 	if it is <= 0.
 	Possible values are [-31, 31], inclusive.
-	Default: 2
+	Default: 1
 
 tcp_allowed_congestion_control - STRING
 	Show/set the congestion control choices available to non-privileged

@@ -407,7 +407,7 @@ tcp_rmem - vector of 3 INTEGERs: min, de
 	net.core.rmem_max.  Calling setsockopt() with SO_RCVBUF disables
 	automatic tuning of that socket's receive buffer size, in which
 	case this value is ignored.
-	Default: between 87380B and 4MB, depending on RAM size.
+	Default: between 87380B and 6MB, depending on RAM size.
 
 tcp_sack - BOOLEAN
 	Enable select acknowledgments (SACKS).

@@ -3221,7 +3221,7 @@ void __init tcp_init(void)
 {
 	struct sk_buff *skb = NULL;
 	unsigned long limit;
-	int i, max_share, cnt;
+	int i, max_rshare, max_wshare, cnt;
 	unsigned long jiffy = jiffies;
 
 	BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb));

@@ -3285,15 +3285,16 @@ void __init tcp_init(void)
 
 	/* Set per-socket limits to no more than 1/128 the pressure threshold */
 	limit = ((unsigned long)sysctl_tcp_mem[1]) << (PAGE_SHIFT - 7);
-	max_share = min(4UL*1024*1024, limit);
+	max_wshare = min(4UL*1024*1024, limit);
+	max_rshare = min(6UL*1024*1024, limit);
 
 	sysctl_tcp_wmem[0] = SK_MEM_QUANTUM;
 	sysctl_tcp_wmem[1] = 16*1024;
-	sysctl_tcp_wmem[2] = max(64*1024, max_share);
+	sysctl_tcp_wmem[2] = max(64*1024, max_wshare);
 
 	sysctl_tcp_rmem[0] = SK_MEM_QUANTUM;
 	sysctl_tcp_rmem[1] = 87380;
-	sysctl_tcp_rmem[2] = max(87380, max_share);
+	sysctl_tcp_rmem[2] = max(87380, max_rshare);
 
 	printk(KERN_INFO "TCP: Hash tables configured "
 	       "(established %u bind %u)\n",

@@ -83,7 +83,7 @@ int sysctl_tcp_ecn __read_mostly = 2;
 EXPORT_SYMBOL(sysctl_tcp_ecn);
 int sysctl_tcp_dsack __read_mostly = 1;
 int sysctl_tcp_app_win __read_mostly = 31;
-int sysctl_tcp_adv_win_scale __read_mostly = 2;
+int sysctl_tcp_adv_win_scale __read_mostly = 1;
 EXPORT_SYMBOL(sysctl_tcp_adv_win_scale);
 
 int sysctl_tcp_stdurg __read_mostly;

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

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

From: Josh Boyer <[email protected]>

commit 6fe6ae56a7cebaebc2e6daa11c423e4692f9b592 upstream.

When the keyboard backlight support was originally added, the commit said to default it to on with a 10 second timeout. That actually wasn't the case, as the default value is commented out for the kbd_backlight parameter. Because it is a static variable, it gets set to 0 by default without some other form of initialization.

However, it seems the function to set the value wasn't actually called immediately, so whatever state the keyboard was in initially would remain. Then commit df410d522410e67660 was introduced during the 2.6.39 timeframe to immediately set whatever value was present (as well as attempt to restore/reset the state on module removal or resume). That seems to have now forced the light off immediately when the module is loaded unless the option kbd_backlight=1 is specified.

Let's enable it by default again (for the first time). This should solve
https://bugzilla.redhat.com/show_bug.cgi?id=728478

---
 drivers/platform/x86/sony-laptop.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -127,7 +127,7 @@ MODULE_PARM_DESC(minor,
 		 "default is -1 (automatic)");
 #endif
 
-static int kbd_backlight;	/* = 1 */
+static int kbd_backlight = 1;
 module_param(kbd_backlight, int, 0444);
 MODULE_PARM_DESC(kbd_backlight,
 		 "set this to 0 to disable keyboard backlight, "

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

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

From: Mark Hills <[email protected]>

commit c914f55f7cdfafe9d7d5b248751902c7ab57691e upstream.

This assertion seems to imply that chip->dsp_code_to_load is a pointer. It's actually an integer handle on the actual firmware, and 0 has no special meaning.

The assertion prevents initialisation of a Darla20 card, but would also affect other models. It seems it was introduced in commit dd7b254d.

ALSA sound/pci/echoaudio/echoaudio.c:2061 Echoaudio driver starting...
ALSA sound/pci/echoaudio/echoaudio.c:1969 chip=ebe4e000
ALSA sound/pci/echoaudio/echoaudio.c:2007 pci=ed568000 irq=19 subdev=0010 Init hardware...
ALSA sound/pci/echoaudio/darla20_dsp.c:36 init_hw() - Darla20
------------[ cut here ]------------
WARNING: at sound/pci/echoaudio/echoaudio_dsp.c:478 init_hw+0x1d1/0x86c [snd_darla20]()
Hardware name: Dell DM051

BUG? (!chip->dsp_code_to_load || !chip->comm_page)

---
 sound/pci/echoaudio/echoaudio_dsp.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -475,7 +475,7 @@ static int load_firmware(struct echoaudi
 	const struct firmware *fw;
 	int box_type, err;
 
-	if (snd_BUG_ON(!chip->dsp_code_to_load || !chip->comm_page))
+	if (snd_BUG_ON(!chip->comm_page))
 		return -EPERM;
 
 	/* See if the ASIC is present and working - only if the DSP is already loaded */

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

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

From: David Henningsson <[email protected]>

commit 32cf4023e689ad5b3a81a749d8cc99d7f184cb99 upstream.

When an IRQ for some reason gets lost, we wait up to a second using udelay, which is CPU intensive. This patch improves the situation by waiting about 30 ms in the CPU intensive mode, then stepping down to using msleep(2) instead. In essence, we trade some granularity in exchange for less CPU consumption when the waiting time is a bit longer.

As a result, PulseAudio should no longer be killed by the kernel for taking up to much RT-prio CPU time. At least not for *this* reason.

---
 sound/pci/hda/hda_intel.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

@@ -702,11 +702,13 @@ static unsigned int azx_rirb_get_respons
 {
 	struct azx *chip = bus->private_data;
 	unsigned long timeout;
+	unsigned long loopcounter;
 	int do_poll = 0;
 
  again:
 	timeout = jiffies + msecs_to_jiffies(1000);
-	for (;;) {
+
+	for (loopcounter = 0;; loopcounter++) {
 		if (chip->polling_mode || do_poll) {
 			spin_lock_irq(&chip->reg_lock);
 			azx_update_rirb(chip);

@@ -722,7 +724,7 @@ static unsigned int azx_rirb_get_respons
 		}
 		if (time_after(jiffies, timeout))
 			break;
-		if (bus->needs_damn_long_delay)
+		if (bus->needs_damn_long_delay || loopcounter > 3000)
 			msleep(2); /* temporary workaround */
 		else {
 			udelay(10);

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

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

From: Ming Lei <[email protected]>

commit 5b6e9bcdeb65634b4ad604eb4536404bbfc62cfa upstream.

Commit 4231d47e6fe69f061f96c98c30eaf9fb4c14b96d(net/usbnet: avoid recursive locking in usbnet_stop()) fixes the recursive locking problem by releasing the skb queue lock before unlink, but may cause skb traversing races:
- after URB is unlinked and the queue lock is released, the refered skb and skb->next may be moved to done queue, even be released
- in skb_queue_walk_safe, the next skb is still obtained by next pointer of the last skb
- so maybe trigger oops or other problems

This patch extends the usage of entry->state to describe 'start_unlink' state, so always holding the queue(rx/tx) lock to change the state if the referd skb is in rx or tx queue because we need to know if the refered urb has been started unlinking in unlink_urbs.

The other part of this patch is based on Huajun's patch:
always traverse from head of the tx/rx queue to get skb which is to be unlinked but not been started unlinking.

---
 drivers/net/usb/usbnet.c   |   54 +++++++++++++++++++++++++++++++--------------
 include/linux/usb/usbnet.h |    3 +-
 2 files changed, 40 insertions(+), 17 deletions(-)

@@ -277,17 +277,32 @@ int usbnet_change_mtu (struct net_device
 }
 EXPORT_SYMBOL_GPL(usbnet_change_mtu);
 
+/* The caller must hold list->lock */
+static void __usbnet_queue_skb(struct sk_buff_head *list,
+			struct sk_buff *newsk, enum skb_state state)
+{
+	struct skb_data *entry = (struct skb_data *) newsk->cb;
+
+	__skb_queue_tail(list, newsk);
+	entry->state = state;
+}
+
 /*-------------------------------------------------------------------------*/
 
 /* some LK 2.4 HCDs oopsed if we freed or resubmitted urbs from
  * completion callbacks.  2.5 should have fixed those bugs...
  */
 
-static void defer_bh(struct usbnet *dev, struct sk_buff *skb, struct sk_buff_head *list)
+static enum skb_state defer_bh(struct usbnet *dev, struct sk_buff *skb,
+		struct sk_buff_head *list, enum skb_state state)
 {
 	unsigned long		flags;
+	enum skb_state 		old_state;
+	struct skb_data *entry = (struct skb_data *) skb->cb;
 
 	spin_lock_irqsave(&list->lock, flags);
+	old_state = entry->state;
+	entry->state = state;
 	__skb_unlink(skb, list);
 	spin_unlock(&list->lock);
 	spin_lock(&dev->done.lock);

@@ -295,6 +310,7 @@ static void defer_bh(struct usbnet *dev,
 	if (dev->done.qlen == 1)
 		tasklet_schedule(&dev->bh);
 	spin_unlock_irqrestore(&dev->done.lock, flags);
+	return old_state;
 }
 
 /* some work can't be done in tasklets, so we use keventd

@@ -335,7 +351,6 @@ static int rx_submit (struct usbnet *dev
 	entry = (struct skb_data *) skb->cb;
 	entry->urb = urb;
 	entry->dev = dev;
-	entry->state = rx_start;
 	entry->length = 0;
 
 	usb_fill_bulk_urb (urb, dev->udev, dev->in,

@@ -367,7 +382,7 @@ static int rx_submit (struct usbnet *dev
 			tasklet_schedule (&dev->bh);
 			break;
 		case 0:
-			__skb_queue_tail (&dev->rxq, skb);
+			__usbnet_queue_skb(&dev->rxq, skb, rx_start);
 		}
 	} else {
 		netif_dbg(dev, ifdown, dev->net, "rx: stopped\n");

@@ -418,16 +433,17 @@ static void rx_complete (struct urb *urb
 	struct skb_data		*entry = (struct skb_data *) skb->cb;
 	struct usbnet		*dev = entry->dev;
 	int			urb_status = urb->status;
+	enum skb_state		state;
 
 	skb_put (skb, urb->actual_length);
-	entry->state = rx_done;
+	state = rx_done;
 	entry->urb = NULL;
 
 	switch (urb_status) {
 	/* success */
 	case 0:
 		if (skb->len < dev->net->hard_header_len) {
-			entry->state = rx_cleanup;
+			state = rx_cleanup;
 			dev->net->stats.rx_errors++;
 			dev->net->stats.rx_length_errors++;
 			netif_dbg(dev, rx_err, dev->net,

@@ -466,7 +482,7 @@ static void rx_complete (struct urb *urb
 				  "rx throttle %d\n", urb_status);
 		}
 block:
-		entry->state = rx_cleanup;
+		state = rx_cleanup;
 		entry->urb = urb;
 		urb = NULL;
 		break;

@@ -477,17 +493,18 @@ block:
 		// FALLTHROUGH
 
 	default:
-		entry->state = rx_cleanup;
+		state = rx_cleanup;
 		dev->net->stats.rx_errors++;
 		netif_dbg(dev, rx_err, dev->net, "rx status %d\n", urb_status);
 		break;
 	}
 
-	defer_bh(dev, skb, &dev->rxq);
+	state = defer_bh(dev, skb, &dev->rxq, state);
 
 	if (urb) {
 		if (netif_running (dev->net) &&
-		    !test_bit (EVENT_RX_HALT, &dev->flags)) {
+		    !test_bit (EVENT_RX_HALT, &dev->flags) &&
+		    state != unlink_start) {
 			rx_submit (dev, urb, GFP_ATOMIC);
 			return;
 		}

@@ -573,16 +590,23 @@ EXPORT_SYMBOL_GPL(usbnet_purge_paused_rx
 static int unlink_urbs (struct usbnet *dev, struct sk_buff_head *q)
 {
 	unsigned long		flags;
-	struct sk_buff		*skb, *skbnext;
+	struct sk_buff		*skb;
 	int			count = 0;
 
 	spin_lock_irqsave (&q->lock, flags);
-	skb_queue_walk_safe(q, skb, skbnext) {
+	while (!skb_queue_empty(q)) {
 		struct skb_data		*entry;
 		struct urb		*urb;
 		int			retval;
 
-		entry = (struct skb_data *) skb->cb;
+		skb_queue_walk(q, skb) {
+			entry = (struct skb_data *) skb->cb;
+			if (entry->state != unlink_start)
+				goto found;
+		}
+		break;
+found:
+		entry->state = unlink_start;
 		urb = entry->urb;
 
 		/*

@@ -1033,8 +1057,7 @@ static void tx_complete (struct urb *urb
 	}
 
 	usb_autopm_put_interface_async(dev->intf);
-	entry->state = tx_done;
-	defer_bh(dev, skb, &dev->txq);
+	(void) defer_bh(dev, skb, &dev->txq, tx_done);
 }
 
 /*-------------------------------------------------------------------------*/

@@ -1087,7 +1110,6 @@ netdev_tx_t usbnet_start_xmit (struct sk
 	entry = (struct skb_data *) skb->cb;
 	entry->urb = urb;
 	entry->dev = dev;
-	entry->state = tx_start;
 	entry->length = length;
 
 	usb_fill_bulk_urb (urb, dev->udev, dev->out,

@@ -1146,7 +1168,7 @@ netdev_tx_t usbnet_start_xmit (struct sk
 		break;
 	case 0:
 		net->trans_start = jiffies;
-		__skb_queue_tail (&dev->txq, skb);
+		__usbnet_queue_skb(&dev->txq, skb, tx_start);
 		if (dev->txq.qlen >= TX_QLEN (dev))
 			netif_stop_queue (net);
 	}

@@ -191,7 +191,8 @@ extern void usbnet_cdc_status(struct usb
 enum skb_state {
 	illegal = 0,
 	tx_start, tx_done,
-	rx_start, rx_done, rx_cleanup
+	rx_start, rx_done, rx_cleanup,
+	unlink_start
 };
 
 struct skb_data {	/* skb->cb is one of these */

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

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

From: Mike Galbraith <[email protected]>

commit 5e2bf0142231194d36fdc9596b36a261ed2b9fe7 upstream.

Fork() failure post namespace creation for a child cloned with CLONE_NEWPID leaks pid_namespace/mnt_cache due to proc being mounted during creation, but not unmounted during cleanup. Call pid_ns_release_proc() during cleanup.

---
 kernel/fork.c |    3 +++
 1 file changed, 3 insertions(+)

@@ -48,6 +48,7 @@
 #include <linux/audit.h>
 #include <linux/memcontrol.h>
 #include <linux/ftrace.h>
+#include <linux/proc_fs.h>
 #include <linux/profile.h>
 #include <linux/rmap.h>
 #include <linux/ksm.h>

@@ -1378,6 +1379,8 @@ bad_fork_cleanup_io:
 	if (p->io_context)
 		exit_io_context(p);
 bad_fork_cleanup_namespaces:
+	if (unlikely(clone_flags & CLONE_NEWPID))
+		pid_ns_release_proc(p->nsproxy->pid_ns);
 	exit_task_namespaces(p);
 bad_fork_cleanup_mm:
 	if (p->mm) {

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

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

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

[ Upstream commit a5a737e090e25981e99d69f01400e3a80356581c ]

%g2 is meant to hold the CPUID number throughout this routine, since at the very beginning, and at the very end, we use %g2 to calculate indexes into per-cpu arrays.

However we erroneously clobber it in order to hold the %cwp register value mid-stream.

Fix this code to use %g3 for the %cwp read and related calulcations instead.

---
 arch/sparc/kernel/central.c |    2 +-
 arch/sparc/mm/ultra.S       |    6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

@@ -268,4 +268,4 @@ static int __init sunfire_init(void)
 	return 0;
 }
 
-subsys_initcall(sunfire_init);
+fs_initcall(sunfire_init);

@@ -495,11 +495,11 @@ xcall_fetch_glob_regs:
 	stx		%o7, [%g1 + GR_SNAP_O7]
 	stx		%i7, [%g1 + GR_SNAP_I7]
 	/* Don't try this at home kids... */
-	rdpr		%cwp, %g2
-	sub		%g2, 1, %g7
+	rdpr		%cwp, %g3
+	sub		%g3, 1, %g7
 	wrpr		%g7, %cwp
 	mov		%i7, %g7
-	wrpr		%g2, %cwp
+	wrpr		%g3, %cwp
 	stx		%g7, [%g1 + GR_SNAP_RPC]
 	sethi		%hi(trap_block), %g7
 	or		%g7, %lo(trap_block), %g7

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

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

From: Russell King <[email protected]>

commit 9b61a4d1b2064dbd0c9e61754305ac852170509f upstream.

---
 arch/arm/mm/fault.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

@@ -266,7 +266,9 @@ good_area:
 	return fault;
 
 check_stack:
-	if (vma->vm_flags & VM_GROWSDOWN && !expand_stack(vma, addr))
+	/* Don't allow expansion below FIRST_USER_ADDRESS */
+	if (vma->vm_flags & VM_GROWSDOWN &&
+	    addr >= FIRST_USER_ADDRESS && !expand_stack(vma, addr))
 		goto good_area;
 out:
 	return fault;

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

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

From: Luis Henriques <[email protected]>

commit 9ef449c6b31bb6a8e6dedc24de475a3b8c79be20 upstream.

An early registration of an ISR was causing a crash to several users (for example, with the ite-cir driver: http://bugs.launchpad.net/bugs/972723).
The reason was that IRQs were being triggered before a driver initialisation was completed.

This patch fixes this by moving the invocation to request_irq() and to request_region() to a later stage on the driver probe function.

---
 drivers/media/rc/ene_ir.c      |   32 ++++++++--------
 drivers/media/rc/fintek-cir.c  |   20 +++++-----
 drivers/media/rc/ite-cir.c     |   20 +++++-----
 drivers/media/rc/nuvoton-cir.c |   36 +++++++++---------
 drivers/media/rc/winbond-cir.c |   78 ++++++++++++++++++++---------------------
 5 files changed, 93 insertions(+), 93 deletions(-)

@@ -1017,22 +1017,6 @@ static int ene_probe(struct pnp_dev *pnp
 
 	spin_lock_init(&dev->hw_lock);
 
-	/* claim the resources */
-	error = -EBUSY;
-	dev->hw_io = pnp_port_start(pnp_dev, 0);
-	if (!request_region(dev->hw_io, ENE_IO_SIZE, ENE_DRIVER_NAME)) {
-		dev->hw_io = -1;
-		dev->irq = -1;
-		goto error;
-	}
-
-	dev->irq = pnp_irq(pnp_dev, 0);
-	if (request_irq(dev->irq, ene_isr,
-			IRQF_SHARED, ENE_DRIVER_NAME, (void *)dev)) {
-		dev->irq = -1;
-		goto error;
-	}
-
 	pnp_set_drvdata(pnp_dev, dev);
 	dev->pnp_dev = pnp_dev;
 

@@ -1085,6 +1069,22 @@ static int ene_probe(struct pnp_dev *pnp
 	device_set_wakeup_capable(&pnp_dev->dev, true);
 	device_set_wakeup_enable(&pnp_dev->dev, true);
 
+	/* claim the resources */
+	error = -EBUSY;
+	dev->hw_io = pnp_port_start(pnp_dev, 0);
+	if (!request_region(dev->hw_io, ENE_IO_SIZE, ENE_DRIVER_NAME)) {
+		dev->hw_io = -1;
+		dev->irq = -1;
+		goto error;
+	}
+
+	dev->irq = pnp_irq(pnp_dev, 0);
+	if (request_irq(dev->irq, ene_isr,
+			IRQF_SHARED, ENE_DRIVER_NAME, (void *)dev)) {
+		dev->irq = -1;
+		goto error;
+	}
+
 	error = rc_register_device(rdev);
 	if (error < 0)
 		goto error;

@@ -504,16 +504,6 @@ static int fintek_probe(struct pnp_dev *
 
 	spin_lock_init(&fintek->fintek_lock);
 
-	ret = -EBUSY;
-	/* now claim resources */
-	if (!request_region(fintek->cir_addr,
-			    fintek->cir_port_len, FINTEK_DRIVER_NAME))
-		goto failure;
-
-	if (request_irq(fintek->cir_irq, fintek_cir_isr, IRQF_SHARED,
-			FINTEK_DRIVER_NAME, (void *)fintek))
-		goto failure;
-
 	pnp_set_drvdata(pdev, fintek);
 	fintek->pdev = pdev;
 

@@ -548,6 +538,16 @@ static int fintek_probe(struct pnp_dev *
 	/* rx resolution is hardwired to 50us atm, 1, 25, 100 also possible */
 	rdev->rx_resolution = US_TO_NS(CIR_SAMPLE_PERIOD);
 
+	ret = -EBUSY;
+	/* now claim resources */
+	if (!request_region(fintek->cir_addr,
+			    fintek->cir_port_len, FINTEK_DRIVER_NAME))
+		goto failure;
+
+	if (request_irq(fintek->cir_irq, fintek_cir_isr, IRQF_SHARED,
+			FINTEK_DRIVER_NAME, (void *)fintek))
+		goto failure;
+
 	ret = rc_register_device(rdev);
 	if (ret)
 		goto failure;

@@ -1519,16 +1519,6 @@ static int ite_probe(struct pnp_dev *pde
 	/* initialize raw event */
 	init_ir_raw_event(&itdev->rawir);
 
-	ret = -EBUSY;
-	/* now claim resources */
-	if (!request_region(itdev->cir_addr,
-				dev_desc->io_region_size, ITE_DRIVER_NAME))
-		goto failure;
-
-	if (request_irq(itdev->cir_irq, ite_cir_isr, IRQF_SHARED,
-			ITE_DRIVER_NAME, (void *)itdev))
-		goto failure;
-
 	/* set driver data into the pnp device */
 	pnp_set_drvdata(pdev, itdev);
 	itdev->pdev = pdev;

@@ -1604,6 +1594,16 @@ static int ite_probe(struct pnp_dev *pde
 	rdev->driver_name = ITE_DRIVER_NAME;
 	rdev->map_name = RC_MAP_RC6_MCE;
 
+	ret = -EBUSY;
+	/* now claim resources */
+	if (!request_region(itdev->cir_addr,
+				dev_desc->io_region_size, ITE_DRIVER_NAME))
+		goto failure;
+
+	if (request_irq(itdev->cir_irq, ite_cir_isr, IRQF_SHARED,
+			ITE_DRIVER_NAME, (void *)itdev))
+		goto failure;
+
 	ret = rc_register_device(rdev);
 	if (ret)
 		goto failure;

@@ -1027,24 +1027,6 @@ static int nvt_probe(struct pnp_dev *pde
 	spin_lock_init(&nvt->nvt_lock);
 	spin_lock_init(&nvt->tx.lock);
 
-	ret = -EBUSY;
-	/* now claim resources */
-	if (!request_region(nvt->cir_addr,
-			    CIR_IOREG_LENGTH, NVT_DRIVER_NAME))
-		goto failure;
-
-	if (request_irq(nvt->cir_irq, nvt_cir_isr, IRQF_SHARED,
-			NVT_DRIVER_NAME, (void *)nvt))
-		goto failure;
-
-	if (!request_region(nvt->cir_wake_addr,
-			    CIR_IOREG_LENGTH, NVT_DRIVER_NAME))
-		goto failure;
-
-	if (request_irq(nvt->cir_wake_irq, nvt_cir_wake_isr, IRQF_SHARED,
-			NVT_DRIVER_NAME, (void *)nvt))
-		goto failure;
-
 	pnp_set_drvdata(pdev, nvt);
 	nvt->pdev = pdev;
 

@@ -1091,6 +1073,24 @@ static int nvt_probe(struct pnp_dev *pde
 	rdev->tx_resolution = XYZ;
 #endif
 
+	ret = -EBUSY;
+	/* now claim resources */
+	if (!request_region(nvt->cir_addr,
+			    CIR_IOREG_LENGTH, NVT_DRIVER_NAME))
+		goto failure;
+
+	if (request_irq(nvt->cir_irq, nvt_cir_isr, IRQF_SHARED,
+			NVT_DRIVER_NAME, (void *)nvt))
+		goto failure;
+
+	if (!request_region(nvt->cir_wake_addr,
+			    CIR_IOREG_LENGTH, NVT_DRIVER_NAME))
+		goto failure;
+
+	if (request_irq(nvt->cir_wake_irq, nvt_cir_wake_isr, IRQF_SHARED,
+			NVT_DRIVER_NAME, (void *)nvt))
+		goto failure;
+
 	ret = rc_register_device(rdev);
 	if (ret)
 		goto failure;

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

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

From: Bjørn Mork <[email protected]>

commit 6eddcb4c82883451aec3be1240f17793370fa62f upstream.

Some RNDIS devices include a bogus CDC Union descriptor pointing to non-existing interfaces. The RNDIS code is already prepared to handle devices without a CDC Union descriptor by hardwiring the driver to use interfaces 0 and 1, which is correct for the devices with the bogus descriptor as well. So we can reuse the existing workaround.

---
 drivers/net/usb/cdc_ether.c |   14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

@@ -83,6 +83,7 @@ int usbnet_generic_cdc_bind(struct usbne
 	struct cdc_state		*info = (void *) &dev->data;
 	int				status;
 	int				rndis;
+	bool				android_rndis_quirk = false;
 	struct usb_driver		*driver = driver_of(intf);
 	struct usb_cdc_mdlm_desc	*desc = NULL;
 	struct usb_cdc_mdlm_detail_desc *detail = NULL;

@@ -195,6 +196,11 @@ int usbnet_generic_cdc_bind(struct usbne
 					info->control,
 					info->u->bSlaveInterface0,
 					info->data);
+				/* fall back to hard-wiring for RNDIS */
+				if (rndis) {
+					android_rndis_quirk = true;
+					goto next_desc;
+				}
 				goto bad_desc;
 			}
 			if (info->control != intf) {

@@ -271,11 +277,15 @@ next_desc:
 	/* Microsoft ActiveSync based and some regular RNDIS devices lack the
 	 * CDC descriptors, so we'll hard-wire the interfaces and not check
 	 * for descriptors.
+	 *
+	 * Some Android RNDIS devices have a CDC Union descriptor pointing
+	 * to non-existing interfaces.  Ignore that and attempt the same
+	 * hard-wired 0 and 1 interfaces.
 	 */
-	if (rndis && !info->u) {
+	if (rndis && (!info->u || android_rndis_quirk)) {
 		info->control = usb_ifnum_to_if(dev->udev, 0);
 		info->data = usb_ifnum_to_if(dev->udev, 1);
-		if (!info->control || !info->data) {
+		if (!info->control || !info->data || info->control != intf) {
 			dev_dbg(&intf->dev,
 				"rndis: master #0/%p slave #1/%p\n",
 				info->control,

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

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

From: Dan Williams <[email protected]>

commit 4e6304b8420aba5311ba21fd68dab2924ae4d91a upstream.

Needs to be tagged with FLAG_WWAN, which since it has generic descriptors, won't happen if we don't override the generic driver info.

---
 drivers/net/usb/cdc_ether.c |   16 ++++++++++++++++
 1 file changed, 16 insertions(+)

@@ -482,6 +482,7 @@ static const struct driver_info wwan_inf
 /*-------------------------------------------------------------------------*/
 
 #define HUAWEI_VENDOR_ID	0x12D1
+#define NOVATEL_VENDOR_ID	0x1410
 
 static const struct usb_device_id	products [] = {
 /*

@@ -599,6 +600,21 @@ static const struct usb_device_id	produc
  * because of bugs/quirks in a given product (like Zaurus, above).
  */
 {
+	/* Novatel USB551L */
+	/* This match must come *before* the generic CDC-ETHER match so that
+	 * we get FLAG_WWAN set on the device, since it's descriptors are
+	 * generic CDC-ETHER.
+	 */
+	.match_flags    =   USB_DEVICE_ID_MATCH_VENDOR
+		 | USB_DEVICE_ID_MATCH_PRODUCT
+		 | USB_DEVICE_ID_MATCH_INT_INFO,
+	.idVendor               = NOVATEL_VENDOR_ID,
+	.idProduct		= 0xB001,
+	.bInterfaceClass	= USB_CLASS_COMM,
+	.bInterfaceSubClass	= USB_CDC_SUBCLASS_ETHERNET,
+	.bInterfaceProtocol	= USB_CDC_PROTO_NONE,
+	.driver_info = (unsigned long)&wwan_info,
+}, {
 	USB_INTERFACE_INFO(USB_CLASS_COMM, USB_CDC_SUBCLASS_ETHERNET,
 			USB_CDC_PROTO_NONE),
 	.driver_info = (unsigned long) &cdc_info,

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

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

From: Tejun Heo <[email protected]>

commit 42b64281453249dac52861f9b97d18552a7ec62b upstream.

pcpu_embed_first_chunk() allocates memory for each node, copies percpu data and frees unused portions of it before proceeding to the next group. This assumes that allocations for different nodes doesn't overlap; however, depending on memory topology, the bootmem allocator may end up allocating memory from a different node than the requested one which may overlap with the portion freed from one of the previous percpu areas. This leads to percpu groups for different nodes overlapping which is a serious bug.

This patch separates out copy & partial free from the allocation loop such that all allocations are complete before partial frees happen.

This also fixes overlapping frees which could happen on allocation failure path - out_free_areas path frees whole groups but the groups could have portions freed at that point.

LKML-Reference: <[email protected]>

---
 mm/percpu.c |   10 ++++++++++
 1 file changed, 10 insertions(+)

@@ -1630,6 +1630,16 @@ int __init pcpu_embed_first_chunk(size_t
 		areas[group] = ptr;
 
 		base = min(ptr, base);
+	}
+
+	/*
+	 * Copy data and free unused parts.  This should happen after all
+	 * allocations are complete; otherwise, we may end up with
+	 * overlapping groups.
+	 */
+	for (group = 0; group < ai->nr_groups; group++) {
+		struct pcpu_group_info *gi = &ai->groups[group];
+		void *ptr = areas[group];
 
 		for (i = 0; i < gi->nr_units; i++, ptr += ai->unit_size) {
 			if (gi->cpu_map[i] == NR_CPUS) {

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

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

From: Catalin Marinas <[email protected]>

commit 100d13c3b5b9410f604b86f5e0a34da64b8cf659 upstream.

Kmemleak tracks the percpu allocations via a specific API and the originally allocated areas must be removed from kmemleak (via kmemleak_free). The code was already doing this for SMP systems.

---
 mm/percpu.c |    2 ++
 1 file changed, 2 insertions(+)

@@ -1873,6 +1873,8 @@ void __init setup_per_cpu_areas(void)
 	fc = __alloc_bootmem(unit_size, PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
 	if (!ai || !fc)
 		panic("Failed to allocate memory for percpu areas.");
+	/* kmemleak tracks the percpu allocations separately */
+	kmemleak_free(fc);
 
 	ai->dyn_size = unit_size;
 	ai->unit_size = unit_size;

Greg KH wrote...

    #include <linux/kmemleak.h>

Odd, it doesn't cause that problem here for me, what ARCH are you building for? I'm on x86-64.

Actually, I don't think this is needed at all for 3.0 as percpu memleak handling went in with the patch you reference above, so this would cause problems on 3.0. So I'll go delete that patch from the queue, thanks for letting me know.

greg k-h

Greg KH wrote...

x86-64 and ARM, not x86-32 for whatever reason, parisc not tested yet.

But, if I read the nested #ifdef right, CONFIG_CMP must be disabled to hit that code.

I'm fine with that descision.

Christoph

This patch propagated to far in the past. Support for per-cpu areas in kmemleak was added with commit f528f0b8e5 (kmemleak: Handle percpu memory allocation) in 3.3-rc1. That patch introduced the correct #include for kmemleak.h.

Commit 100d13c should only be applied to stable 3.3 and not earlier.

Thanks.

Catalin

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

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

From: Chris Metcalf <[email protected]>

commit 4998a6c0edce7fae9c0a5463f6ec3fa585258ee7 upstream.

Commit 66aebce747eaf ("hugetlb: fix race condition in hugetlb_fault()") added code to avoid a race condition by elevating the page refcount in hugetlb_fault() while calling hugetlb_cow().

However, one code path in hugetlb_cow() includes an assertion that the page count is 1, whereas it may now also have the value 2 in this path.

The consensus is that this BUG_ON has served its purpose, so rather than extending it to cover both cases, we just remove it.

---
 mm/hugetlb.c |    1 -
 1 file changed, 1 deletion(-)

@@ -2398,7 +2398,6 @@ retry_avoidcopy:
 		if (outside_reserve) {
 			BUG_ON(huge_pte_none(pte));
 			if (unmap_ref_private(mm, vma, old_page, address)) {
-				BUG_ON(page_count(old_page) != 1);
 				BUG_ON(huge_pte_none(pte));
 				spin_lock(&mm->page_table_lock);
 				goto retry_avoidcopy;

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

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

From: Russ Anderson <[email protected]>

commit 6bc2e853c6b46a6041980d58200ad9b0a73a60ff upstream.

Systems with 8 TBytes of memory or greater can hit a problem where only the the first 8 TB of memory shows up. This is due to "int i" being smaller than "unsigned long start_aligned", causing the high bits to be dropped.

The fix is to change `i' to unsigned long to match start_aligned and end_aligned.

Thanks to Jack Steiner for assistance tracking this down.

---
 mm/nobootmem.c |    3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

@@ -83,8 +83,7 @@ void __init free_bootmem_late(unsigned l
 
 static void __init __free_pages_memory(unsigned long start, unsigned long end)
 {
-	int i;
-	unsigned long start_aligned, end_aligned;
+	unsigned long i, start_aligned, end_aligned;
 	int order = ilog2(BITS_PER_LONG);
 
 	start_aligned = (start + (BITS_PER_LONG - 1)) & ~(BITS_PER_LONG - 1);

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

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

From: Josh Cartwright <[email protected]>

commit 226bb7df3d22bcf4a1c0fe8206c80cc427498eae upstream.

The locking policy is such that the erase_complete_block spinlock is nested within the alloc_sem mutex. This fixes a case in which the acquisition order was erroneously reversed. This issue was caught by the following lockdep splat:

=======================================================
[ INFO: possible circular locking dependency detected ] 3.0.5 #1
-------------------------------------------------------
jffs2_gcd_mtd6/299 is trying to acquire lock:

    (&c->alloc_sem){+.+.+.}, at: [<c01f7714>] jffs2_garbage_collect_pass+0x314/0x890

but task is already holding lock:

    (&(&c->erase_completion_lock)->rlock){+.+...}, at: [<c01f7708>] jffs2_garbage_collect_pass+0x308/0x890

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #1 (&(&c->erase_completion_lock)->rlock){+.+...}:

          [<c008bec4>] validate_chain+0xe6c/0x10bc
          [<c008c660>] __lock_acquire+0x54c/0xba4
          [<c008d240>] lock_acquire+0xa4/0x114
          [<c046780c>] _raw_spin_lock+0x3c/0x4c
          [<c01f744c>] jffs2_garbage_collect_pass+0x4c/0x890
          [<c01f937c>] jffs2_garbage_collect_thread+0x1b4/0x1cc
          [<c0071a68>] kthread+0x98/0xa0
          [<c000f264>] kernel_thread_exit+0x0/0x8

-> #0 (&c->alloc_sem){+.+.+.}:

          [<c008ad2c>] print_circular_bug+0x70/0x2c4
          [<c008c08c>] validate_chain+0x1034/0x10bc
          [<c008c660>] __lock_acquire+0x54c/0xba4
          [<c008d240>] lock_acquire+0xa4/0x114
          [<c0466628>] mutex_lock_nested+0x74/0x33c
          [<c01f7714>] jffs2_garbage_collect_pass+0x314/0x890
          [<c01f937c>] jffs2_garbage_collect_thread+0x1b4/0x1cc
          [<c0071a68>] kthread+0x98/0xa0
          [<c000f264>] kernel_thread_exit+0x0/0x8

other info that might help us debug this:

    Possible unsafe locking scenario:

          CPU0                    CPU1
          ----                    ----
     lock(&(&c->erase_completion_lock)->rlock);
                                  lock(&c->alloc_sem);
                                  lock(&(&c->erase_completion_lock)->rlock);
     lock(&c->alloc_sem);

    *** DEADLOCK ***

1 lock held by jffs2_gcd_mtd6/299:

    #0:  (&(&c->erase_completion_lock)->rlock){+.+...}, at: [<c01f7708>] jffs2_garbage_collect_pass+0x308/0x890

   stack backtrace:
   [<c00155dc>] (unwind_backtrace+0x0/0x100) from [<c0463dc0>] (dump_stack+0x20/0x24)
   [<c0463dc0>] (dump_stack+0x20/0x24) from [<c008ae84>] (print_circular_bug+0x1c8/0x2c4)
   [<c008ae84>] (print_circular_bug+0x1c8/0x2c4) from [<c008c08c>] (validate_chain+0x1034/0x10bc)
   [<c008c08c>] (validate_chain+0x1034/0x10bc) from [<c008c660>] (__lock_acquire+0x54c/0xba4)
   [<c008c660>] (__lock_acquire+0x54c/0xba4) from [<c008d240>] (lock_acquire+0xa4/0x114)
   [<c008d240>] (lock_acquire+0xa4/0x114) from [<c0466628>] (mutex_lock_nested+0x74/0x33c)
   [<c0466628>] (mutex_lock_nested+0x74/0x33c) from [<c01f7714>] (jffs2_garbage_collect_pass+0x314/0x890)
   [<c01f7714>] (jffs2_garbage_collect_pass+0x314/0x890) from [<c01f937c>] (jffs2_garbage_collect_thread+0x1b4/0x1cc)
   [<c01f937c>] (jffs2_garbage_collect_thread+0x1b4/0x1cc) from [<c0071a68>] (kthread+0x98/0xa0)
   [<c0071a68>] (kthread+0x98/0xa0) from [<c000f264>] (kernel_thread_exit+0x0/0x8)

This was introduce in '81cfc9f jffs2: Fix serious write stall due to erase'.

---
 fs/jffs2/gc.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -225,8 +225,8 @@ int jffs2_garbage_collect_pass(struct jf
 			return 0;
 
 		D1(printk(KERN_DEBUG "No progress from erasing blocks; doing GC anyway\n"));
-		spin_lock(&c->erase_completion_lock);
 		mutex_lock(&c->alloc_sem);
+		spin_lock(&c->erase_completion_lock);
 	}
 
 	/* First, work out which block we're garbage-collecting */

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

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

From: Chris Metcalf <[email protected]>

commit a134d228298c6aa9007205c6b81cae0cac0acb5d upstream.

This passes siginfo and mcontext to tilegx32 signal handlers that don't have SA_SIGINFO set just as we have been doing for tilegx64.

---
 arch/tile/kernel/compat_signal.c |   12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

@@ -406,19 +406,17 @@ int compat_setup_rt_frame(int sig, struc
 	 * Set up registers for signal handler.
 	 * Registers that we don't modify keep the value they had from
 	 * user-space at the time we took the signal.
+	 * We always pass siginfo and mcontext, regardless of SA_SIGINFO,
+	 * since some things rely on this (e.g. glibc's debug/segfault.c).
 	 */
 	regs->pc = ptr_to_compat_reg(ka->sa.sa_handler);
 	regs->ex1 = PL_ICS_EX1(USER_PL, 1); /* set crit sec in handler */
 	regs->sp = ptr_to_compat_reg(frame);
 	regs->lr = restorer;
 	regs->regs[0] = (unsigned long) usig;
-
-	if (ka->sa.sa_flags & SA_SIGINFO) {
-		/* Need extra arguments, so mark to restore caller-saves. */
-		regs->regs[1] = ptr_to_compat_reg(&frame->info);
-		regs->regs[2] = ptr_to_compat_reg(&frame->uc);
-		regs->flags |= PT_FLAGS_CALLER_SAVES;
-	}
+	regs->regs[1] = ptr_to_compat_reg(&frame->info);
+	regs->regs[2] = ptr_to_compat_reg(&frame->uc);
+	regs->flags |= PT_FLAGS_CALLER_SAVES;
 
 	/*
 	 * Notify any tracer that was single-stepping it.

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

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

From: Alexander Clouter <[email protected]>

commit 1ebfefcf37a6e308266a8d786e8cfea0a454058c upstream.

Without CRYPTO_HASH being selected, mv_cesa has a lot of hooks into undefined exports.
----
MODPOST 81 modules
Kernel: arch/arm/boot/Image is ready

  AS      arch/arm/boot/compressed/head.o
  GZIP    arch/arm/boot/compressed/piggy.gzip
  CC      arch/arm/boot/compressed/misc.o
  CC      arch/arm/boot/compressed/decompress.o

ERROR: "crypto_ahash_type" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_shash_final" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_register_ahash" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_unregister_ahash" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_shash_update" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_shash_digest" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_shash_setkey" [drivers/crypto/mv_cesa.ko] undefined!
ERROR: "crypto_alloc_shash" [drivers/crypto/mv_cesa.ko] undefined! make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2
make: *** Waiting for unfinished jobs....
----

---
 drivers/crypto/Kconfig |    1 +
 1 file changed, 1 insertion(+)

@@ -173,6 +173,7 @@ config CRYPTO_DEV_MV_CESA
 	select CRYPTO_ALGAPI
 	select CRYPTO_AES
 	select CRYPTO_BLKCIPHER2
+	select CRYPTO_HASH
 	help
 	  This driver allows you to utilize the Cryptographic Engines and
 	  Security Accelerator (CESA) which can be found on the Marvell Orion

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

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

From: Jonathan Brassow <[email protected]>

commit 0d9f4f135eb6dea06bdcb7065b1e4ff78274a5e9 upstream.

Use del_timer_sync to remove timer before mddev_suspend finishes.

We don't want a timer going off after an mddev_suspend is called. This is especially true with device-mapper, since it can call the destructor function immediately following a suspend. This results in the removal (kfree) of the structures upon which the timer depends - resulting in a very ugly panic. Therefore, we add a del_timer_sync to mddev_suspend to prevent this.

---
 drivers/md/md.c |    2 ++
 1 file changed, 2 insertions(+)

@@ -348,6 +348,8 @@ void mddev_suspend(mddev_t *mddev)
 	synchronize_rcu();
 	wait_event(mddev->sb_wait, atomic_read(&mddev->active_io) == 0);
 	mddev->pers->quiesce(mddev, 1);
+
+	del_timer_sync(&mddev->safemode_timer);
 }
 EXPORT_SYMBOL_GPL(mddev_suspend);
 

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

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

From: Willy Tarreau <[email protected]>

commit bad115cfe5b509043b684d3a007ab54b80090aa1 upstream.

Since recent changes on TCP splicing (starting with commits 2f533844 "tcp: allow splice() to build full TSO packets" and 35f9c09f "tcp:
tcp_sendpages() should call tcp_push() once"), I started seeing massive stalls when forwarding traffic between two sockets using splice() when pipe buffers were larger than socket buffers.

Latest changes (net: netdev_alloc_skb() use build_skb()) made the problem even more apparent.

The reason seems to be that if do_tcp_sendpages() fails on out of memory condition without being able to send at least one byte, tcp_push() is not called and the buffers cannot be flushed.

After applying the attached patch, I cannot reproduce the stalls at all and the data rate it perfectly stable and steady under any condition which previously caused the problem to be permanent.

The issue seems to have been there since before the kernel migrated to git, which makes me think that the stalls I occasionally experienced with tux during stress-tests years ago were probably related to the same issue.

This issue was first encountered on 3.0.31 and 3.2.17, so please backport to -stable.

---
 net/ipv4/tcp.c |    3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

@@ -850,8 +850,7 @@ new_segment:
 wait_for_sndbuf:
 		set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
 wait_for_memory:
-		if (copied)
-			tcp_push(sk, flags & ~MSG_MORE, mss_now, TCP_NAGLE_PUSH);
+		tcp_push(sk, flags & ~MSG_MORE, mss_now, TCP_NAGLE_PUSH);
 
 		if ((err = sk_stream_wait_memory(sk, &timeo)) != 0)
 			goto do_error;

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

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

From: Sasha Levin <[email protected]>

commit 377485f6244af255b04d662cf19cddbbc4ae4310 upstream.

Currently, we'll try mounting any device who's major device number is UNNAMED_MAJOR as NFS root. This would happen for non-NFS devices as well (such as 9p devices) but it wouldn't cause any issues since mounting the device as NFS would fail quickly and the code proceeded to doing the proper mount:

       [  101.522716] VFS: Unable to mount root fs via NFS, trying floppy.
       [  101.534499] VFS: Mounted root (9p filesystem) on device 0:18.

Commit 6829a048102a ("NFS: Retry mounting NFSROOT") introduced retries when mounting NFS root, which means that now we don't immediately fail and instead it takes an additional 90+ seconds until we stop retrying, which has revealed the issue this patch fixes.

This meant that it would take an additional 90 seconds to boot when we're not using a device type which gets detected in order before NFS.

This patch modifies the NFS type check to require device type to be 'Root_NFS' instead of requiring the device to have an UNNAMED_MAJOR major. This makes boot process cleaner since we now won't go through the NFS mounting code at all when the device isn't an NFS root ("/dev/nfs").

---
 init/do_mounts.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -432,7 +432,7 @@ void __init change_floppy(char *fmt, ...
 void __init mount_root(void)
 {
 #ifdef CONFIG_ROOT_NFS
-	if (MAJOR(ROOT_DEV) == UNNAMED_MAJOR) {
+	if (ROOT_DEV == Root_NFS) {
 		if (mount_nfs_root())
 			return;
 

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

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

From: Eric Sandeen <[email protected]>

commit c1bb05a657fb3d8c6179a4ef7980261fae4521d7 upstream.

Processes hang forever on a sync-mounted ext2 file system that is mounted with the ext4 module (default in Fedora 16).

I can reproduce this reliably by mounting an ext2 partition with "-o sync" and opening a new file an that partition with vim. vim will hang in "D" state forever. The same happens on ext4 without a journal.

I am attaching a small patch here that solves this issue for me. In the sync mounted case without a journal, ext4_handle_dirty_metadata() may call sync_dirty_buffer(), which can't be called with buffer lock held.

Also move mb_cache_entry_release inside lock to avoid race fixed previously by 8a2bfdcb ext[34]: EA block reference count racing fix Note too that ext2 fixed this same problem in 2006 with b2f49033 [PATCH] fix deadlock in ext2

---
 fs/ext4/xattr.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

@@ -487,18 +487,19 @@ ext4_xattr_release_block(handle_t *handl
 		ext4_free_blocks(handle, inode, bh, 0, 1,
 				 EXT4_FREE_BLOCKS_METADATA |
 				 EXT4_FREE_BLOCKS_FORGET);
+		unlock_buffer(bh);
 	} else {
 		le32_add_cpu(&BHDR(bh)->h_refcount, -1);
+		if (ce)
+			mb_cache_entry_release(ce);
+		unlock_buffer(bh);
 		error = ext4_handle_dirty_metadata(handle, inode, bh);
 		if (IS_SYNC(inode))
 			ext4_handle_sync(handle);
 		dquot_free_block(inode, 1);
 		ea_bdebug(bh, "refcount now=%d; releasing",
 			  le32_to_cpu(BHDR(bh)->h_refcount));
-		if (ce)
-			mb_cache_entry_release(ce);
 	}
-	unlock_buffer(bh);
 out:
 	ext4_std_error(inode->i_sb, error);
 	return;

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

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

From: Jonathan Nieder <[email protected]>

This is a shorter (and more appropriate for stable kernels) analog to the following upstream commit:

commit 6926afd1925a54a13684ebe05987868890665e2b

Author: Trond Myklebust <[email protected]>
Date:   Sat Jan 7 13:22:46 2012 -0500

NFSv4: Save the owner/group name string when doing open

...so that we can do the uid/gid mapping outside the asynchronous RPC context.
This fixes a bug in the current NFSv4 atomic open code where the client isn't able to determine what the true uid/gid fields of the file are, (because the asynchronous nature of the OPEN call denies it the ability to do an upcall) and so fills them with default values, marking the inode as needing revalidation.
Unfortunately, in some cases, the VFS will do some additional sanity checks on the file, and may override the server's decision to allow the open because it sees the wrong owner/group fields.

Without this patch, logging into two different machines with home directories mounted over NFS4 and then running "vim" and typing ":q" in each reliably produces the following error on the second machine:

E137: Viminfo file is not writable: /users/system/rtheys/.viminfo

This regression was introduced by 80e52aced138 ("NFSv4: Don't do idmapper upcalls for asynchronous RPC calls", merged during the 2.6.32 cycle) --- after the OPEN call, .viminfo has the default values for st_uid and st_gid (0xfffffffe) cached because we do not want to let rpciod wait for an idmapper upcall to fill them in.

The fix used in mainline is to save the owner and group as strings and perform the upcall in _nfs4_proc_open outside the rpciod context, which takes about 600 lines. For stable, we can do something similar with a one-liner: make open check for the stale fields and make a (synchronous) GETATTR call to fill them when needed.

Trond dictated the patch, I typed it in, and Rik tested it.

Addresses http://bugs.debian.org/659111 and

          https://bugzilla.redhat.com/789298

---
 fs/nfs/nfs4proc.c |    1 +
 1 file changed, 1 insertion(+)

@@ -1771,6 +1771,7 @@ static int _nfs4_do_open(struct inode *d
 			nfs_setattr_update_inode(state->inode, sattr);
 		nfs_post_op_update_inode(state->inode, opendata->o_res.f_attr);
 	}
+	nfs_revalidate_inode(server, state->inode);
 	nfs4_opendata_put(opendata);
 	nfs4_put_state_owner(sp);
 	*res = state;

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

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

From: Sha Zhengju <[email protected]>

commit 8c7577637ca31385e92769a77e2ab5b428e8b99c upstream.

When the last event is unregistered, there is no need to keep the spare array anymore. So free it to avoid memory leak.

---
 mm/memcontrol.c |    6 ++++++
 1 file changed, 6 insertions(+)

@@ -4605,6 +4605,12 @@ static void mem_cgroup_usage_unregister_
 swap_buffers:
 	/* Swap primary and spare array */
 	thresholds->spare = thresholds->primary;
+	/* If all events are unregistered, free the spare array */
+	if (!new) {
+		kfree(thresholds->spare);
+		thresholds->spare = NULL;
+	}
+
 	rcu_assign_pointer(thresholds->primary, new);
 
 	/* To be sure that nobody uses thresholds */

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

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

From: Jan Kiszka <[email protected]>

commit b7dafa0ef3145c31d7753be0a08b3cbda51f0209 upstream.

compat_sys_sigprocmask reads a smaller signal mask from userspace than sigprogmask accepts for setting. So the high word of blocked.sig[0] will be cleared, releasing any potentially blocked RT signal.

This was discovered via userspace code that relies on get/setcontext. glibc's i386 versions of those functions use sigprogmask instead of rt_sigprogmask to save/restore signal mask and caused RT signal unblocking this way.

As suggested by Linus, this replaces the sys_sigprocmask based compat version with one that open-codes the required logic, including the merge of the existing blocked set with the new one provided on SIG_SETMASK.

---
 kernel/compat.c |   65 ++++++++++++++++++++++++++++++++++++++++----------------
 1 file changed, 47 insertions(+), 18 deletions(-)

@@ -318,25 +318,54 @@ asmlinkage long compat_sys_sigpending(co
 
 #ifdef __ARCH_WANT_SYS_SIGPROCMASK
 
-asmlinkage long compat_sys_sigprocmask(int how, compat_old_sigset_t __user *set,
-		compat_old_sigset_t __user *oset)
+/*
+ * sys_sigprocmask SIG_SETMASK sets the first (compat) word of the
+ * blocked set of signals to the supplied signal set
+ */
+static inline void compat_sig_setmask(sigset_t *blocked, compat_sigset_word set)
 {
-	old_sigset_t s;
-	long ret;
-	mm_segment_t old_fs;
-
-	if (set && get_user(s, set))
-		return -EFAULT;
-	old_fs = get_fs();
-	set_fs(KERNEL_DS);
-	ret = sys_sigprocmask(how,
-			      set ? (old_sigset_t __user *) &s : NULL,
-			      oset ? (old_sigset_t __user *) &s : NULL);
-	set_fs(old_fs);
-	if (ret == 0)
-		if (oset)
-			ret = put_user(s, oset);
-	return ret;
+	memcpy(blocked->sig, &set, sizeof(set));
+}
+
+asmlinkage long compat_sys_sigprocmask(int how,
+				       compat_old_sigset_t __user *nset,
+				       compat_old_sigset_t __user *oset)
+{
+	old_sigset_t old_set, new_set;
+	sigset_t new_blocked;
+
+	old_set = current->blocked.sig[0];
+
+	if (nset) {
+		if (get_user(new_set, nset))
+			return -EFAULT;
+		new_set &= ~(sigmask(SIGKILL) | sigmask(SIGSTOP));
+
+		new_blocked = current->blocked;
+
+		switch (how) {
+		case SIG_BLOCK:
+			sigaddsetmask(&new_blocked, new_set);
+			break;
+		case SIG_UNBLOCK:
+			sigdelsetmask(&new_blocked, new_set);
+			break;
+		case SIG_SETMASK:
+			compat_sig_setmask(&new_blocked, new_set);
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		set_current_blocked(&new_blocked);
+	}
+
+	if (oset) {
+		if (put_user(old_set, oset))
+			return -EFAULT;
+	}
+
+	return 0;
 }
 
 #endif

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

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

From: Jan Kara <[email protected]>

commit 1415dd8705394399d59a3df1ab48d149e1e41e77 upstream.

When insert_inode_locked() fails in ext3_new_inode() it most likely means inode bitmap got corrupted and we allocated again inode which is already in use. Also doing unlock_new_inode() during error recovery is wrong since inode does not have I_NEW set. Fix the problem by jumping to fail: (instead of fail_drop:) which declares filesystem error and does not call unlock_new_inode().

---
 fs/ext3/ialloc.c |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

@@ -561,8 +561,12 @@ got:
 	if (IS_DIRSYNC(inode))
 		handle->h_sync = 1;
 	if (insert_inode_locked(inode) < 0) {
-		err = -EINVAL;
-		goto fail_drop;
+		/*
+		 * Likely a bitmap corruption causing inode to be allocated
+		 * twice.
+		 */
+		err = -EIO;
+		goto fail;
 	}
 	spin_lock(&sbi->s_next_gen_lock);
 	inode->i_generation = sbi->s_next_generation++;

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

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

From: Jan Kara <[email protected]>

commit acd6ad83517639e8f09a8c5525b1dccd81cd2a10 upstream.

When insert_inode_locked() fails in ext4_new_inode() it most likely means inode bitmap got corrupted and we allocated again inode which is already in use. Also doing unlock_new_inode() during error recovery is wrong since the inode does not have I_NEW set. Fix the problem by jumping to fail: (instead of fail_drop:) which declares filesystem error and does not call unlock_new_inode().

---
 fs/ext4/ialloc.c |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

@@ -1021,8 +1021,12 @@ got:
 	if (IS_DIRSYNC(inode))
 		ext4_handle_sync(handle);
 	if (insert_inode_locked(inode) < 0) {
-		err = -EINVAL;
-		goto fail_drop;
+		/*
+		 * Likely a bitmap corruption causing inode to be allocated
+		 * twice.
+		 */
+		err = -EIO;
+		goto fail;
 	}
 	spin_lock(&sbi->s_next_gen_lock);
 	inode->i_generation = sbi->s_next_generation++;

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

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

From: Lan Tianyu <[email protected]>

commit 93f770846e8dedc5d9117bd4ad9d7efd18420627 upstream.

Sony Vaio VPCCW29FX does not resume correctly without acpi_sleep=nonvs, so add it to the ACPI sleep blacklist.

https://bugzilla.kernel.org/show_bug.cgi?id=34722

---
 drivers/acpi/sleep.c |    8 ++++++++
 1 file changed, 8 insertions(+)

@@ -422,6 +422,14 @@ static struct dmi_system_id __initdata a
 	},
 	{
 	.callback = init_nvs_nosave,
+	.ident = "Sony Vaio VPCCW29FX",
+	.matches = {
+		DMI_MATCH(DMI_SYS_VENDOR, "Sony Corporation"),
+		DMI_MATCH(DMI_PRODUCT_NAME, "VPCCW29FX"),
+		},
+	},
+	{
+	.callback = init_nvs_nosave,
 	.ident = "Averatec AV1020-ED2",
 	.matches = {
 		DMI_MATCH(DMI_SYS_VENDOR, "AVERATEC"),

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

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

From: "Stephen M. Cameron" <[email protected]>

commit 45bcf018d1a4779d592764ef57517c92589d55d7 upstream.

IRQF_SHARED is required for older controllers that don't support MSI(X) and which may end up sharing an interrupt. All the controllers hpsa normally supports have MSI(X) capability, but older controllers may be encountered via the hpsa_allow_any=1 module parameter.

Also remove deprecated IRQF_DISABLED.

---
 drivers/scsi/hpsa.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

@@ -4037,10 +4037,10 @@ static int hpsa_request_irq(struct ctlr_
 
 	if (h->msix_vector || h->msi_vector)
 		rc = request_irq(h->intr[h->intr_mode], msixhandler,
-				IRQF_DISABLED, h->devname, h);
+				0, h->devname, h);
 	else
 		rc = request_irq(h->intr[h->intr_mode], intxhandler,
-				IRQF_DISABLED, h->devname, h);
+				IRQF_SHARED, h->devname, h);
 	if (rc) {
 		dev_err(&h->pdev->dev, "unable to get irq %d for %s\n",
 		       h->intr[h->intr_mode], h->devname);

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

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

From: Kazuya Mio <[email protected]>

commit e1616300a20c80396109c1cf013ba9a36055a3da upstream.

dd slept infinitely when fsfeeze failed because of EIO. To fix this problem, if ->freeze_fs fails, freeze_super() wakes up the tasks waiting for the filesystem to become unfrozen.

When s_frozen isn't SB_UNFROZEN in __generic_file_aio_write(), the function sleeps until FITHAW ioctl wakes up s_wait_unfrozen.

However, if ->freeze_fs fails, s_frozen is set to SB_UNFROZEN and then freeze_super() returns an error number. In this case, FITHAW ioctl returns EINVAL because s_frozen is already SB_UNFROZEN. There is no way to wake up s_wait_unfrozen, so __generic_file_aio_write() sleeps infinitely.

---
 fs/super.c |    2 ++
 1 file changed, 2 insertions(+)

@@ -1009,6 +1009,8 @@ int freeze_super(struct super_block *sb)
 			printk(KERN_ERR
 				"VFS:Filesystem freeze failed\n");
 			sb->s_frozen = SB_UNFROZEN;
+			smp_wmb();
+			wake_up(&sb->s_wait_unfrozen);
 			deactivate_locked_super(sb);
 			return ret;
 		}

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

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

From: Tomoya MORINAGA <[email protected]>

commit 868fea0507308b6548bba7debe5f5c2d5ca47fca upstream.

ML7831 is companion chip for Intel Atom E6xx series.

---
 drivers/gpio/Kconfig    |   11 ++++++-----
 drivers/gpio/pch_gpio.c |    1 +
 2 files changed, 7 insertions(+), 5 deletions(-)

@@ -350,18 +350,19 @@ config GPIO_LANGWELL
 	  Say Y here to support Intel Langwell/Penwell GPIO.
 
 config GPIO_PCH
-	tristate "Intel EG20T PCH / OKI SEMICONDUCTOR ML7223 IOH GPIO"
+	tristate "Intel EG20T PCH/LAPIS Semiconductor IOH(ML7223/ML7831) GPIO"
 	depends on PCI && X86
 	help
 	  This driver is for PCH(Platform controller Hub) GPIO of Intel Topcliff
 	  which is an IOH(Input/Output Hub) for x86 embedded processor.
 	  This driver can access PCH GPIO device.
 
-	  This driver also can be used for OKI SEMICONDUCTOR IOH(Input/
-	  Output Hub), ML7223.
+	  This driver also can be used for LAPIS Semiconductor IOH(Input/
+	  Output Hub), ML7223 and ML7831.
 	  ML7223 IOH is for MP(Media Phone) use.
-	  ML7223 is companion chip for Intel Atom E6xx series.
-	  ML7223 is completely compatible for Intel EG20T PCH.
+	  ML7831 IOH is for general purpose use.
+	  ML7223/ML7831 is companion chip for Intel Atom E6xx series.
+	  ML7223/ML7831 is completely compatible for Intel EG20T PCH.
 
 config GPIO_ML_IOH
 	tristate "OKI SEMICONDUCTOR ML7213 IOH GPIO support"

@@ -287,6 +287,7 @@ static int pch_gpio_resume(struct pci_de
 static DEFINE_PCI_DEVICE_TABLE(pch_gpio_pcidev_id) = {
 	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x8803) },
 	{ PCI_DEVICE(PCI_VENDOR_ID_ROHM, 0x8014) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_ROHM, 0x8803) },
 	{ 0, }
 };
 MODULE_DEVICE_TABLE(pci, pch_gpio_pcidev_id);

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

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

From: Toshiharu Okada <[email protected]>

commit 5229d87edcd80a3bceb0708ebd767faff2e589a9 upstream.

This patch fixed the issue which receives an unnecessary packet before link

When using PHY of GMII, an unnecessary packet is received, And it becomes impossible to receive a packet after link up.

---
 drivers/net/pch_gbe/pch_gbe_main.c |   21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

@@ -717,13 +717,6 @@ static void pch_gbe_configure_rx(struct
 	iowrite32(rdba, &hw->reg->RX_DSC_BASE);
 	iowrite32(rdlen, &hw->reg->RX_DSC_SIZE);
 	iowrite32((rdba + rdlen), &hw->reg->RX_DSC_SW_P);
-
-	/* Enables Receive DMA */
-	rxdma = ioread32(&hw->reg->DMA_CTRL);
-	rxdma |= PCH_GBE_RX_DMA_EN;
-	iowrite32(rxdma, &hw->reg->DMA_CTRL);
-	/* Enables Receive */
-	iowrite32(PCH_GBE_MRE_MAC_RX_EN, &hw->reg->MAC_RX_EN);
 }
 
 /**

@@ -1097,6 +1090,19 @@ void pch_gbe_update_stats(struct pch_gbe
 	spin_unlock_irqrestore(&adapter->stats_lock, flags);
 }
 
+static void pch_gbe_start_receive(struct pch_gbe_hw *hw)
+{
+	u32 rxdma;
+
+	/* Enables Receive DMA */
+	rxdma = ioread32(&hw->reg->DMA_CTRL);
+	rxdma |= PCH_GBE_RX_DMA_EN;
+	iowrite32(rxdma, &hw->reg->DMA_CTRL);
+	/* Enables Receive */
+	iowrite32(PCH_GBE_MRE_MAC_RX_EN, &hw->reg->MAC_RX_EN);
+	return;
+}
+
 /**
  * pch_gbe_intr - Interrupt Handler
  * @irq:   Interrupt number

@@ -1717,6 +1723,7 @@ int pch_gbe_up(struct pch_gbe_adapter *a
 	pch_gbe_alloc_tx_buffers(adapter, tx_ring);
 	pch_gbe_alloc_rx_buffers(adapter, rx_ring, rx_ring->count);
 	adapter->tx_queue_len = netdev->tx_queue_len;
+	pch_gbe_start_receive(&adapter->hw);
 
 	mod_timer(&adapter->watchdog_timer, jiffies);
 

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

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

From: Toshiharu Okada <[email protected]>

commit 7756332f5b64c9c1535712b9679792e8bd4f0019 upstream.

Support new device OKI SEMICONDUCTOR ML7831 IOH(Input/Output Hub)

ML7831 is for general purpose use.
ML7831 is companion chip for Intel Atom E6xx series. ML7831 is completely compatible for Intel EG20T PCH.

---
 drivers/net/Kconfig                |   11 ++++++-----
 drivers/net/pch_gbe/pch_gbe_main.c |   10 ++++++++++
 2 files changed, 16 insertions(+), 5 deletions(-)

@@ -2543,7 +2543,7 @@ config S6GMAC
 source "drivers/net/stmmac/Kconfig"
 
 config PCH_GBE
-	tristate "Intel EG20T PCH / OKI SEMICONDUCTOR ML7223 IOH GbE"
+	tristate "Intel EG20T PCH/OKI SEMICONDUCTOR IOH(ML7223/ML7831) GbE"
 	depends on PCI
 	select MII
 	---help---

@@ -2556,10 +2556,11 @@ config PCH_GBE
 	  This driver enables Gigabit Ethernet function.
 
 	  This driver also can be used for OKI SEMICONDUCTOR IOH(Input/
-	  Output Hub), ML7223.
-	  ML7223 IOH is for MP(Media Phone) use.
-	  ML7223 is companion chip for Intel Atom E6xx series.
-	  ML7223 is completely compatible for Intel EG20T PCH.
+	  Output Hub), ML7223/ML7831.
+	  ML7223 IOH is for MP(Media Phone) use. ML7831 IOH is for general
+	  purpose use.
+	  ML7223/ML7831 is companion chip for Intel Atom E6xx series.
+	  ML7223/ML7831 is completely compatible for Intel EG20T PCH.
 
 endif # NETDEV_1000
 

@@ -39,6 +39,9 @@ const char pch_driver_version[] = DRV_VE
 #define PCI_VENDOR_ID_ROHM			0x10db
 #define PCI_DEVICE_ID_ROHM_ML7223_GBE		0x8013
 
+/* Macros for ML7831 */
+#define PCI_DEVICE_ID_ROHM_ML7831_GBE		0x8802
+
 #define PCH_GBE_TX_WEIGHT         64
 #define PCH_GBE_RX_WEIGHT         64
 #define PCH_GBE_RX_BUFFER_WRITE   16

@@ -2457,6 +2460,13 @@ static DEFINE_PCI_DEVICE_TABLE(pch_gbe_p
 	 .subvendor = PCI_ANY_ID,
 	 .subdevice = PCI_ANY_ID,
 	 .class = (PCI_CLASS_NETWORK_ETHERNET << 8),
+	 .class_mask = (0xFFFF00)
+	 },
+	{.vendor = PCI_VENDOR_ID_ROHM,
+	 .device = PCI_DEVICE_ID_ROHM_ML7831_GBE,
+	 .subvendor = PCI_ANY_ID,
+	 .subdevice = PCI_ANY_ID,
+	 .class = (PCI_CLASS_NETWORK_ETHERNET << 8),
 	 .class_mask = (0xFFFF00)
 	 },
 	/* required last entry */

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

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

From: Toshiharu Okada <[email protected]>

commit 5f3a11419099d5cc010cfbfc524ca10d8fb81f89 upstream.

When a link was downed during network use,
there is an issue on which PC freezes.

This patch fixed this issue.

---
 drivers/net/pch_gbe/pch_gbe_main.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -2128,7 +2128,7 @@ static int pch_gbe_napi_poll(struct napi
 		/* If no Tx and not enough Rx work done,
 		 * exit the polling mode
 		 */
-		if ((work_done < budget) || !netif_running(netdev))
+		if (work_done < budget)
 			poll_end_flag = true;
 	}
 

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

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

From: Darren Hart <[email protected]>

commit 2b53d07891630dead46d65c8f896955fd3ae0302 upstream.

If the MAC is invalid or not implemented, do not abort the probe. Issue a warning and prevent bringing the interface up until a MAC is set manually (via ifconfig $IFACE hw ether $MAC).

Tested on two platforms, one with a valid MAC, the other without a MAC. The real MAC is used if present, the interface fails to come up until the MAC is set on the other. They successfully get an IP over DHCP and pass a simple ping and login over ssh test.

This is meant to allow the Inforce SYS940X development board:
http://www.inforcecomputing.com/SYS940X_ECX.html
(and others suffering from a missing MAC) to work with the mainline kernel. Without this patch, the probe will fail and the interface will not be created, preventing the user from configuring the MAC manually.

This does not make any attempt to address a missing or invalid MAC for the pch_phub driver.

---
 drivers/net/pch_gbe/pch_gbe_main.c |   17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

@@ -1710,6 +1710,12 @@ int pch_gbe_up(struct pch_gbe_adapter *a
 	struct pch_gbe_rx_ring *rx_ring = adapter->rx_ring;
 	int err;
 
+	/* Ensure we have a valid MAC */
+	if (!is_valid_ether_addr(adapter->hw.mac.addr)) {
+		pr_err("Error: Invalid MAC address\n");
+		return -EINVAL;
+	}
+
 	/* hardware has been reset, we need to reload some things */
 	pch_gbe_set_multi(netdev);
 

@@ -2402,9 +2408,14 @@ static int pch_gbe_probe(struct pci_dev
 
 	memcpy(netdev->dev_addr, adapter->hw.mac.addr, netdev->addr_len);
 	if (!is_valid_ether_addr(netdev->dev_addr)) {
-		dev_err(&pdev->dev, "Invalid MAC Address\n");
-		ret = -EIO;
-		goto err_free_adapter;
+		/*
+		 * If the MAC is invalid (or just missing), display a warning
+		 * but do not abort setting up the device. pch_gbe_up will
+		 * prevent the interface from being brought up until a valid MAC
+		 * is set.
+		 */
+		dev_err(&pdev->dev, "Invalid MAC address, "
+		                    "interface disabled.\n");
 	}
 	setup_timer(&adapter->watchdog_timer, pch_gbe_watchdog,
 		    (unsigned long)adapter);

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

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

From: Dan Carpenter <[email protected]>

commit 73f98eab9b9e0bab492ca06add5657d9e702ddb1 upstream.

pch_gbe_validate_option() modifies 32 bits of memory but we pass &hw->phy.autoneg_advertised which only has 16 bits and &hw->mac.fc which only has 8 bits.

---
 drivers/net/pch_gbe/pch_gbe_param.c |   15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

@@ -320,10 +320,10 @@ static void pch_gbe_check_copper_options
 			pr_debug("AutoNeg specified along with Speed or Duplex, AutoNeg parameter ignored\n");
 			hw->phy.autoneg_advertised = opt.def;
 		} else {
-			hw->phy.autoneg_advertised = AutoNeg;
-			pch_gbe_validate_option(
-				(int *)(&hw->phy.autoneg_advertised),
-				&opt, adapter);
+			int tmp = AutoNeg;
+
+			pch_gbe_validate_option(&tmp, &opt, adapter);
+			hw->phy.autoneg_advertised = tmp;
 		}
 	}
 

@@ -494,9 +494,10 @@ void pch_gbe_check_options(struct pch_gb
 			.arg  = { .l = { .nr = (int)ARRAY_SIZE(fc_list),
 					 .p = fc_list } }
 		};
-		hw->mac.fc = FlowControl;
-		pch_gbe_validate_option((int *)(&hw->mac.fc),
-						&opt, adapter);
+		int tmp = FlowControl;
+
+		pch_gbe_validate_option(&tmp, &opt, adapter);
+		hw->mac.fc = tmp;
 	}
 
 	pch_gbe_check_copper_options(adapter);

You really want people to work over the weekends don't you? ;-)

Well, I set up a box to run ktest randconfigs and some other checks. Right now it's running on the 3.2.18-rc release. If I can sneak away from the misses, I may be able to kick off tests against these releases too. But I only have one box dedicated for this, so everything needs to be processed serially.

-- Steve

Sorry :)

I will wait until late on Monday before doing the real release, so you do get an extra day here.

No rush, do it on Monday if needed. Thanks for testing, I appreciate it.

greg k-h

Ran 18 tests against 3.0.32-rc, 2 failed (16 passed!). One failed by a lock up (still looking into it) and the other is this one:

Calibrating delay loop (skipped), value calculated using timer frequency.. 4799.59 BogoMIPS (lpj=2399799)
pid_max: default: 32768 minimum: 301

BUG: scheduling while atomic: swapper/0/0x10000002
Modules linked in:
Pid: 0, comm: swapper Not tainted 3.1.0-rc9-test+ #1
Call Trace:
 [<ffffffff81435b3e>] __schedule_bug+0x57/0x5b
 [<ffffffff8143c44f>] __schedule+0x90/0x672
 [<ffffffff810636dd>] ? kzalloc.constprop.2+0x24/0x26
 [<ffffffff810460d1>] __cond_resched+0x23/0x2f
 [<ffffffff8143ca83>] _cond_resched+0x14/0x1d^M
 [<ffffffff810f7cd3>] slab_pre_alloc_hook.isra.52+0x28/0x2c
 [<ffffffff810f8b1a>] kmem_cache_alloc_trace+0x29/0xbd
 [<ffffffff810636dd>] kzalloc.constprop.2+0x24/0x26
 [<ffffffff81b7a434>] pidmap_init+0x81/0xc0
 [<ffffffff81b61b03>] start_kernel+0x330/0x3c2
 [<ffffffff81b612c4>] x86_64_start_reservations+0xaf/0xb3
 [<ffffffff81b61140>] ? early_idt_handlers+0x140/0x140
 [<ffffffff81b613ca>] x86_64_start_kernel+0x102/0x111

Security Framework initialized

Had CONFIG_SLUB, IRQ_FORCED_THREADING and voluntary preempt enabled. Seems the might_resched() check from might_sleep_if() in pre_alloc_hook() caused a schedule in early boot (pidmap_init). Not sure why it scheduled that early, perhaps a threaded interrupt was added? I haven't been able to reproduce it again.

Perhaps we should not be using GFP_KERNEL when allocating memory for pidmap_init()? As we have preemption disabled at this point, we should be using GFP_ATOMIC. Also add a BUG_ON if the page fails to get allocated.

This patch is against 3.4-rc2 (just happened to be a kernel checkout I had).

index 9f08dfa..32dab20 100644
@@ -563,7 +563,9 @@ void __init pidmap_init(void)
 				PIDS_PER_CPU_MIN * num_possible_cpus());
 	pr_info("pid_max: default: %u minimum: %u\n", pid_max, pid_max_min);
 
-	init_pid_ns.pidmap[0].page = kzalloc(PAGE_SIZE, GFP_KERNEL);
+	/* Preemption is still disabled at this point */
+	init_pid_ns.pidmap[0].page = kzalloc(PAGE_SIZE, GFP_ATOMIC);
+	BUG_ON(!init_pid_ns.pidmap[0].page);
 	/* Reserve PID 0. We never call free_pidmap(0) */
 	set_bit(0, init_pid_ns.pidmap[0].page);
 	atomic_dec(&init_pid_ns.pidmap[0].nr_free);

We set gfp_allowed_mask to allow all allocations before this point: it happens when we enable interrupts fairly early during start_kernel().

So by the time pidmap_init() is called, GFP_KERNEL does imply that scheduling can happen.

Which does imply that we set gfp_allowed_mask *much* too early. We still cannot schedule at that point (well, at least there's a comment saying so):

        /*
         * Disable preemption - early bootup scheduling is extremely
         * fragile until we cpu_idle() for the first time.
         */
        preempt_disable();

so logically we should move the gfp_allowed_mask setting down to where we really are properly alive.

How about moving it down to after we've done the full smp_init() and after we've actually done the first schedule and have proper idle CPU's?

Something like the attached (UNTESTED!) patch?

Linus

 init/main.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

index 44b2433334c7..0cf00943d755 100644
@@ -560,9 +560,6 @@ asmlinkage void __init start_kernel(void)
 	early_boot_irqs_disabled = false;
 	local_irq_enable();
 
-	/* Interrupts are enabled now so all GFP allocations are safe. */
-	gfp_allowed_mask = __GFP_BITS_MASK;
-
 	kmem_cache_init_late();
 
 	/*

@@ -861,6 +858,9 @@ static int __init kernel_init(void * unused)
 	smp_init();
 	sched_init_smp();
 
+	/* Now we're finally set up and can do any allocation */
+	gfp_allowed_mask = __GFP_BITS_MASK;
+
 	do_basic_setup();
 
 	/* Open the /dev/console on the rootfs, this should never fail */

Steven notes that he's using CONFIG_PREEMPT_VOLUNTARY which means all might_sleep()'s in the kernel actually get turned into cond_resched()'s. I had thought that might_sleep() and might_sleep_if() such as in the slab allocators were simply for CONFIG_DEBUG_ATOMIC_SLEEP, but it has been hijacked for other purposes. Do we really want to infuse cond_resched()'s on every slab allocation for desktop users who are urged in the Kconfig to enable CONFIG_PREEMPT_VOLUNTARY?

Well, yes, we can pretty much rely on scheduling having to work at the top of kernel_init(), since kernel_init is being run in a new thread (and thus must have scheduled from the original thread that becomes the first idle thread).

But I thought we might as well delay it until the system was really up, since I wasn't entirely sure what the heck the other CPU's might be doing. That said, I don't really care deeply, I think that anywhere in kernel_init should be fine. I have no strong opinions, but the current location does seem buggy.

That said, I'm not going to delay 3.4 over this (in fact, I already tagged it locally, but haven't pushed out yet because the back-end kernel.org machines seem to be unreachable right now). It apparently only matters for configurations that are not really all that realistic.

Linus