-stable review
Subsystems (11)
cover -stable review
This is the start of the stable review cycle for the 2.6.11.12 release. There are 9 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let us know. If anyone is a maintainer of the proper subsystem, and wants to add a
signed-off-by: line to the patch, please respond with it.
These patches are sent out with a number of different people on the Cc: line.
If you wish to be a reviewer, please email [email protected] to add your name tothe list. If you want to be off the reviewer list, also email us.
Responses should be made by Fri, Jun 10, 23:00 UTC. Anything received after that time, might be too late.
thanks,
the -stable release team (i.e. the ones wearing the joker hat in the corner...)
PATCH 1/9 try_to_unmap_cluster() passes out-of-bounds pte to pte_unmap()
try_to_unmap_cluster() does:
for (pte = pte_offset_map(pmd, address);
address < end; pte++, address += PAGE_SIZE) {
...
}
pte_unmap(pte);It may take a little staring to notice, but pte can actually fall off the end of the pte page in this iteration, which makes life difficult for kmap_atomic() and the users not expecting it to BUG(). Of course, we're somewhat lucky in that arithmetic elsewhere in the function guarantees that at least one iteration is made, lest this force larger rearrangements to be made. This issue and patch also apply to non-mm mainline and with trivial adjustments, at least two related kernels.
Discovered during internal testing at Oracle.
mm/rmap.c +3 -3
@@ -641,7 +641,7 @@pgd_t *pgd;pud_t *pud;pmd_t *pmd; -pte_t *pte; +pte_t *pte, *original_pte;pte_t pteval;struct page *page;unsigned long address;
@@ -673,7 +673,7 @@if (!pmd_present (*pmd)) goto out_unlock; -for (pte =pte_offset_map (pmd, address); +for (original_pte = pte =pte_offset_map (pmd, address); address < end; pte++, address +=PAGE_SIZE ) {if (!pte_present (*pte))
@@ -710,7 +710,7 @@ (*mapcount)--; } -pte_unmap (pte); +pte_unmap (original_pte); out_unlock:spin_unlock (&mm->page_table_lock);
PATCH 2/9 : Fix deadlock with ip_queue and tcp local input path
In reply to Chris WrightWhen we have ip_queue being used from LOCAL_IN, then we end up with a situation where the verdicts coming back from userspace traverse the TCP input path from syscall context. While this seems to work most of the time, there's an ugly deadlock:
syscall context is interrupted by the timer interrupt. When the timer interrupt leaves, the timer softirq get's scheduled and calls tcp_delack_timer() and alike. They themselves do bh_lock_sock(sk), which is already held from somewhere else -> boom.
I've now tested the suggested solution by Patrick McHardy and Herbert Xu to simply use local_bh_{en,dis}able().
net/ipv4/netfilter/ip_queue.c +10 -0
@@ -3,6 +3,7 @@* communicating with userspace via netlink. * * (C) 2000-2002 James Morris <[email protected]> +* (C) 2003-2005 Netfilter Core Team <[email protected]> * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as
@@ -14,6 +15,7 @@* Zander). * 2000-08-01: Added Nick Williams' MAC support. * 2002-06-25: Code cleanup. +* 2005-05-26: local_bh_{disable,enable} around nf_reinject (Harald Welte) * */ #include <linux/module.h>
@@ -66,7 +68,15 @@static void ipq_issue_verdict (struct ipq_queue_entry *entry,int verdict) { +/* TCP input path (and probably other bits) assume to be called +* from softirq context, not from syscall, like ipq_issue_verdict is +* called. TCP input path deadlocks with locks taken from timer +* softirq, e.g. We therefore emulate this by local_bh_disable() */ + +local_bh_disable ();nf_reinject (entry->skb, entry->info, verdict); +local_bh_enable (); +kfree (entry); }
PATCH 3/9 fix hfsplus oops, hfs and hfsplus leak
In reply to Chris WrightThis patch fixes the leak of sb->s_fs_info in both the HFS and HFS+ modules. In addition to this, it fixes an oops happening when trying to mount a non-hfsplus filesystem using hfsplus. This patch is from Roman Zippel, based off patches sent by myself. It's been included in 2.6.12- rc4. See
http://www.kernel.org/git/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=945b092011c6af71a0107be96e119c8c08776f3f
(chrisw: backport to -stable)
fs/hfs/mdb.c | 5 +++++ fs/hfs/super.c | 8 +++----- fs/hfsplus/super.c | 6 +++++- 3 files changed, 13 insertions(+), 6 deletions(-)
fs/hfs/mdb.c +5 -0
@@ -333,6 +333,8 @@ void hfs_mdb_close(struct super_block *s* Release the resources associated with the in-core MDB. */ void hfs_mdb_put (struct super_block *sb) { +if (!HFS_SB (sb)) +return ;/* free the B-trees */ hfs_btree_close (HFS_SB (sb)->ext_tree);hfs_btree_close (HFS_SB (sb)->cat_tree);
@@ -340,4 +342,7 @@ void hfs_mdb_put(struct super_block *sb)/* free the buffers holding the primary and alternate MDBs */ brelse (HFS_SB (sb)->mdb_bh);brelse (HFS_SB (sb)->alt_mdb_bh); + +kfree (HFS_SB (sb)); + sb->s_fs_info =NULL ; }
fs/hfs/super.c +3 -5
@@ -263,7 +263,7 @@ static int hfs_fill_super(struct super_b res = -EINVAL ;if (!parse_options ((char *)data, sbi)) {hfs_warn ("hfs_fs: unable to parse mount options.\n" ); - goto bail3; + goto bail; } sb->s_op = &hfs_super_operations;
@@ -276,7 +276,7 @@ static int hfs_fill_super(struct super_bhfs_warn ("VFS: Can't find a HFS filesystem on dev %s.\n" ,hfs_mdb_name (sb)); res = -EINVAL ; - goto bail2; + goto bail; }/* try to get the root inode */
@@ -306,10 +306,8 @@ bail_iput:iput (root_inode); bail_no_root:hfs_warn ("hfs_fs: get root inode failed.\n" ); +bail:hfs_mdb_put (sb); -bail2: -bail3: -kfree (sbi);return res; }
fs/hfsplus/super.c +5 -1
@@ -207,7 +207,9 @@ static void hfsplus_write_super(struct sstatic void hfsplus_put_super (struct super_block *sb) {dprint (DBG_SUPER ,"hfsplus_put_super\n" ); -if (!(sb->s_flags &MS_RDONLY )) { +if (!sb->s_fs_info) +return ; +if (!(sb->s_flags &MS_RDONLY ) &&HFSPLUS_SB (sb).s_vhdr) {struct hfsplus_vh *vhdr =HFSPLUS_SB (sb).s_vhdr; vhdr->modify_date =hfsp_now2mt ();
@@ -223,6 +225,8 @@ static void hfsplus_put_super(struct supiput (HFSPLUS_SB (sb).alloc_file);iput (HFSPLUS_SB (sb).hidden_dir);brelse (HFSPLUS_SB (sb).s_vhbh); +kfree (sb->s_fs_info); + sb->s_fs_info =NULL ; }static int hfsplus_statfs (struct super_block *sb,struct kstatfs *buf)
PATCH 4/9 x86_64: avoid SMP boot up race
In reply to Chris WrightKeep interrupts disabled during smp bootup
This avoids a race that breaks SMP bootup on some machines. The race is not fully plugged (that is only done with much more changes in 2.6.12), but should be good enough for most people.
Keeping the interrupts disabled here is ok because we don't rely on the timer interrupt for local APIC timer setup, but always read the timer registers directly.
(originally from Rusty Russell iirc)
arch/x86_64/kernel/apic.c renamed from arch/x86_64/kernel/apic.c-o +0 -2
@@ -775,9 +775,7 @@void __init setup_secondary_APIC_clock (void ) { -local_irq_disable ();/* FIXME: Do we need this? --RR */ setup_APIC_timer (calibration_result); -local_irq_enable (); }void __init disable_APIC_timer (void )
arch/x86_64/kernel/smpboot.c renamed from arch/x86_64/kernel/smpboot.c-o +0 -4
@@ -309,8 +309,6 @@Dprintk ("CALLIN, before setup_local_APIC().\n" );setup_local_APIC (); -local_irq_enable (); -/* * Get our bogomips. */
@@ -324,8 +322,6 @@*/ smp_store_cpu_info (cpuid); -local_irq_disable (); -/* * Allow the master to continue. */
PATCH 6/9 Fix for bttv driver (v0.9.15) for Leadtek WinFast VC100 XP capture cards
In reply to Chris WrightThis is a tiny patch that fixes bttv-cards.c so that Leadtek WinFast VC100 XP video capture cards work. I've been advised to post it here after having already posted it to the v4l mailing list.
drivers/media/video/bttv-cards.c renamed from drivers/media/video/bttv-cards.c.orig +0 -1
@@ -1939,7 +1939,6 @@ .no_tda9875 = 1, .no_tda7432 = 1, .tuner_type = TUNER_ABSENT, - .no_video = 1, .pll = PLL_28, },{ .name = "Teppro TEV-560/InterVision IV-560",
PATCH 7/9 ext3: fix log_do_checkpoint() assertion failure
In reply to Chris WrightFix possible false assertion failure in log_do_checkpoint(). We might fail to detect that we actually made a progress when cleaning up the checkpoint lists if we don't retry after writing something to disk. The patch was confirmed to fix observed assertion failures for several users.
When we flushed some buffers we need to retry scanning the list. Otherwise we can fail to detect our progress.
fs/jbd/checkpoint.c +3 -1
@@ -339,8 +339,10 @@ int log_do_checkpoint(journal_t *journal } }while (jh != last_jh && !retry); -if (batch_count) +if (batch_count) {__flush_batch (journal, bhs, &batch_count); + retry =1 ; + }/* * If someone cleaned up this transaction while we slept, we're
PATCH 5/9 x86_64: Fix ptrace boundary check
In reply to Chris WrightDon't allow accesses below register frame in ptrace
There was a "off by one quad word" error in there.
Found and fixed by John Blackwood
arch/x86_64/kernel/ptrace.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)
arch/x86_64/kernel/ptrace.c +2 -2
@@ -252,7 +252,7 @@ asmlinkage long sys_ptrace(long request,break ;switch (addr) { - case 0 ... sizeof(struct user_regs_struct): + case 0 ... sizeof(struct user_regs_struct) - sizeof(long): tmp =getreg (child, addr);break ; case offsetof(struct user, u_debugreg[0]):
@@ -297,7 +297,7 @@ asmlinkage long sys_ptrace(long request,break ;switch (addr) { - case 0 ... sizeof(struct user_regs_struct): + case 0 ... sizeof(struct user_regs_struct) - sizeof(long): ret =putreg (child, addr, data);break ;/* Disallows to set a breakpoint into the vsyscall */
PATCH 8/9 : prevent bad forwarding table updates
In reply to Chris WrightAvoid poisoning of the bridge forwarding table by frames that have been dropped by filtering. This prevents spoofed source addresses on hostile side of bridge from causing packet leakage, a small but possible security risk.
net/bridge/br_input.c +4 -2
@@ -54,6 +54,9 @@ int br_handle_frame_finish(struct sk_bufstruct net_bridge_fdb_entry *dst;int passedup =0 ; +/* insert into forwarding database after filtering to avoid spoofing */ +br_fdb_insert (p->br, p,eth_hdr (skb)->h_source,0 ); +if (br->dev->flags &IFF_PROMISC ) {struct sk_buff *skb2;
@@ -108,8 +111,7 @@ int br_handle_frame(struct net_bridge_poif (eth_hdr (skb)->h_source[0 ] &1 ) goto err; -if (p->state ==BR_STATE_LEARNING || - p->state ==BR_STATE_FORWARDING ) +if (p->state ==BR_STATE_LEARNING )br_fdb_insert (p->br, p,eth_hdr (skb)->h_source,0 ); if (p->br->stp_enabled &&
net/bridge/br_stp_bpdu.c +3 -0
@@ -140,6 +140,9 @@ int br_stp_handle_bpdu(struct sk_buff *sstruct net_bridge *br = p->br;unsigned char *buf; +/* insert into forwarding database after filtering to avoid spoofing */ +br_fdb_insert (p->br, p,eth_hdr (skb)->h_source,0 ); +/* need at least the 802 and STP headers */ if (!pskb_may_pull (skb,sizeof (header)+1 ) ||memcmp (skb->data, header,sizeof (header)))
PATCH 9/9 : netem: duplication fix
In reply to Chris WrightNetem duplication can cause infinite loop in qdisc_run because the qlen of the parent qdisc is not affected by the duplication.
net/sched/sch_netem.c +9 -4
@@ -184,10 +184,15 @@ static int netem_enqueue(struct sk_buff/* Random duplication */ if (q->duplicate && q->duplicate >=get_crandom (&q->dup_cor)) {struct sk_buff *skb2 =skb_clone (skb,GFP_ATOMIC ); - -pr_debug ("netem_enqueue: dup %p\n" , skb2); -if (skb2) -delay_skb (sch, skb2); +if (skb2) { +struct Qdisc *rootq = sch->dev->qdisc; +u32 dupsave = q->duplicate; + +/* prevent duplicating a dup... */ + q->duplicate =0 ; + rootq->enqueue (skb2, rootq); + q->duplicate = dupsave; + } }/* If doing simple delay then gap == 0 so all packets
Re: [patch 04/09] x86_64: avoid SMP boot up race
In reply to Chris WrightHi,
I'm wondering if anyone could comment on if this could be conceived as a security bug. My initial instinct was yes, but on further considertation I can't conceive a way it could be exploited, well not by anyone who couldn't DoS the box in any number of other ways, including shutting it down.
Quoted 58 lines · Chris Wright
Keep interrupts disabled during smp bootup
This avoids a race that breaks SMP bootup on some machines. The race is not fully plugged (that is only done with much more changes in 2.6.12), but should be good enough for most people.
Keeping the interrupts disabled here is ok because we don't rely on the timer interrupt for local APIC timer setup, but always read the timer registers directly.
(originally from Rusty Russell iirc)
Signed-off-by [email protected]Signed-off-by Chris Wright <[email protected]>diff -u linux/arch/x86_64/kernel/apic.c-o
linux/arch/x86_64/kernel/apic.carch/x86_64/kernel/apic.c renamed from arch/x86_64/kernel/apic.c-o +0
-2@@ -775,9 +775,7 @@void __init setup_secondary_APIC_clock (void ) { -local_irq_disable ();/* FIXME: Do we need this? --RR */ setup_APIC_timer (calibration_result); -local_irq_enable (); }void __init disable_APIC_timer (void )diff -u linux/arch/x86_64/kernel/smpboot.c-o
linux-2.6.11/arch/x86_64/kernel/smpboot.carch/x86_64/kernel/smpboot.c renamed from arch/x86_64/kernel/smpboot.c-o +0
-5@@ -309,8 +309,6 @@Dprintk ("CALLIN, before setup_local_APIC().\n" );setup_local_APIC (); -local_irq_enable (); -/* * Get our bogomips. */ @@ -324,8 +322,6 @@*/ smp_store_cpu_info (cpuid); -local_irq_disable (); -/* * Allow the master to continue. */ -To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Signature · 1 line
Horms
Re: [patch 04/09] x86_64: avoid SMP boot up race
Quoted 6 lines · * Horms ([email protected]) wrote:
I'm wondering if anyone could comment on if this could be conceived as a security bug. My initial instinct was yes, but on further considertation I can't conceive a way it could be exploited, well not by anyone who couldn't DoS the box in any number of other ways, including shutting it down.
That code is run early during boot up when bringing online a cpu. If you can control this, you own the box.