From: phil Date: Sat, 17 Apr 2004 05:05:06 +0000 (+0000) Subject: 2.4.21-suse-171 patches and series X-Git-Tag: v1_8_0_110~486^5~250 X-Git-Url: https://git.whamcloud.com/?a=commitdiff_plain;h=8dd07d9cd21c2124a28a8b45b5995d80b45e415b;p=fs%2Flustre-release.git 2.4.21-suse-171 patches and series --- diff --git a/lustre/kernel_patches/patches/configurable-x86-stack-2.4.21-suse-171.patch b/lustre/kernel_patches/patches/configurable-x86-stack-2.4.21-suse-171.patch new file mode 100644 index 0000000..7d2c397 --- /dev/null +++ b/lustre/kernel_patches/patches/configurable-x86-stack-2.4.21-suse-171.patch @@ -0,0 +1,329 @@ +Index: linux-2.4.21-171/arch/i386/kernel/entry.S +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/kernel/entry.S 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/kernel/entry.S 2004-04-03 16:02:32.000000000 -0500 +@@ -45,6 +45,7 @@ + #include + #include + #include ++#include + + EBX = 0x00 + ECX = 0x04 +@@ -130,10 +131,6 @@ + .long 3b,6b; \ + .previous + +-#define GET_CURRENT(reg) \ +- movl $-8192, reg; \ +- andl %esp, reg +- + ENTRY(lcall7) + pushfl # We get a different stack layout with call gates, + pushl %eax # which has to be cleaned up later.. +@@ -149,7 +146,7 @@ + movl %ecx,CS(%esp) # + movl %esp,%ebx + pushl %ebx +- andl $-8192,%ebx # GET_CURRENT ++ andl $-THREAD_SIZE,%ebx # GET_CURRENT + movl exec_domain(%ebx),%edx # Get the execution domain + movl 4(%edx),%edx # Get the lcall7 handler for the domain + pushl $0x7 +@@ -173,7 +170,7 @@ + movl %ecx,CS(%esp) # + movl %esp,%ebx + pushl %ebx +- andl $-8192,%ebx # GET_CURRENT ++ andl $-THREAD_SIZE,%ebx # GET_CURRENT + movl exec_domain(%ebx),%edx # Get the execution domain + movl 4(%edx),%edx # Get the lcall7 handler for the domain + pushl $0x27 +Index: linux-2.4.21-171/arch/i386/kernel/smpboot.c +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/kernel/smpboot.c 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/kernel/smpboot.c 2004-04-03 16:02:32.000000000 -0500 +@@ -833,7 +833,7 @@ + + /* So we see what's up */ + printk("Booting processor %d/%d eip %lx\n", cpu, apicid, start_eip); +- stack_start.esp = (void *) (1024 + PAGE_SIZE + (char *)idle); ++ stack_start.esp = (void *)idle->thread.esp; + + /* + * This grunge runs the startup process for +@@ -914,7 +914,7 @@ + Dprintk("CPU has booted.\n"); + } else { + boot_error= 1; +- if (*((volatile unsigned char *)phys_to_virt(8192)) ++ if (*((volatile unsigned char *)phys_to_virt(THREAD_SIZE)) + == 0xA5) + /* trampoline started but...? */ + printk("Stuck ??\n"); +@@ -937,7 +937,7 @@ + } + + /* mark "stuck" area as not stuck */ +- *((volatile unsigned long *)phys_to_virt(8192)) = 0; ++ *((volatile unsigned long *)phys_to_virt(THREAD_SIZE)) = 0; + + #ifdef CONFIG_ES7000 + if (!es7000_plat) +Index: linux-2.4.21-171/arch/i386/kernel/traps.c +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/kernel/traps.c 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/kernel/traps.c 2004-04-03 16:02:32.000000000 -0500 +@@ -193,7 +193,7 @@ + unsigned long esp = tsk->thread.esp; + + /* User space on another CPU? */ +- if ((esp ^ (unsigned long)tsk) & (PAGE_MASK<<1)) ++ if ((esp ^ (unsigned long)tsk) & ~(THREAD_SIZE - 1)) + return; + show_trace((unsigned long *)esp); + } +Index: linux-2.4.21-171/arch/i386/kernel/head.S +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/kernel/head.S 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/kernel/head.S 2004-04-03 16:02:32.000000000 -0500 +@@ -15,6 +15,7 @@ + #include + #include + #include ++#include + + #define OLD_CL_MAGIC_ADDR 0x90020 + #define OLD_CL_MAGIC 0xA33F +@@ -326,7 +327,7 @@ + ret + + ENTRY(stack_start) +- .long SYMBOL_NAME(init_task_union)+8192 ++ .long SYMBOL_NAME(init_task_union)+THREAD_SIZE + .long __KERNEL_DS + + /* This is the default interrupt "handler" :-) */ +Index: linux-2.4.21-171/arch/i386/kernel/irq.c +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/kernel/irq.c 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/kernel/irq.c 2004-04-03 16:05:17.000000000 -0500 +@@ -597,7 +597,9 @@ + long esp; + + /* Debugging check for stack overflow: is there less than 1KB free? */ +- __asm__ __volatile__("andl %%esp,%0" : "=r" (esp) : "0" (8191)); ++ __asm__ __volatile__( ++ "andl %%esp,%0" ++ : "=r" (esp) : "0" (THREAD_SIZE-1)); + if (unlikely(esp < (sizeof(struct task_struct) + sysctl_stackwarn))) { + static unsigned long next_jiffies; /* ratelimiting */ + static long least_esp = THREAD_SIZE; +Index: linux-2.4.21-171/arch/i386/lib/getuser.S +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/lib/getuser.S 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/lib/getuser.S 2004-04-03 16:02:32.000000000 -0500 +@@ -21,6 +21,10 @@ + * as they get called from within inline assembly. + */ + ++/* Duplicated from asm/processor.h */ ++#include ++#include ++ + addr_limit = 12 + + .text +@@ -28,7 +32,7 @@ + .globl __get_user_1 + __get_user_1: + movl %esp,%edx +- andl $0xffffe000,%edx ++ andl $~(THREAD_SIZE - 1),%edx + cmpl addr_limit(%edx),%eax + jae bad_get_user + 1: movzbl (%eax),%edx +@@ -41,7 +45,7 @@ + addl $1,%eax + movl %esp,%edx + jc bad_get_user +- andl $0xffffe000,%edx ++ andl $~(THREAD_SIZE - 1),%edx + cmpl addr_limit(%edx),%eax + jae bad_get_user + 2: movzwl -1(%eax),%edx +@@ -54,7 +58,7 @@ + addl $3,%eax + movl %esp,%edx + jc bad_get_user +- andl $0xffffe000,%edx ++ andl $~(THREAD_SIZE - 1),%edx + cmpl addr_limit(%edx),%eax + jae bad_get_user + 3: movl -3(%eax),%edx +Index: linux-2.4.21-171/arch/i386/config.in +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/config.in 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/config.in 2004-04-03 16:02:32.000000000 -0500 +@@ -325,6 +325,29 @@ + if [ "$CONFIG_SMP" = "y" -a "$CONFIG_X86_CMPXCHG" = "y" ]; then + define_bool CONFIG_HAVE_DEC_LOCK y + fi ++ ++choice 'Bigger Stack Size Support' \ ++ "off CONFIG_NOBIGSTACK \ ++ 16KB CONFIG_STACK_SIZE_16KB \ ++ 32KB CONFIG_STACK_SIZE_32KB \ ++ 64KB CONFIG_STACK_SIZE_64KB" off ++ ++if [ "$CONFIG_NOBIGSTACK" = "y" ]; then ++ define_int CONFIG_STACK_SIZE_SHIFT 1 ++else ++ if [ "$CONFIG_STACK_SIZE_16KB" = "y" ]; then ++ define_int CONFIG_STACK_SIZE_SHIFT 2 ++ else ++ if [ "$CONFIG_STACK_SIZE_32KB" = "y" ]; then ++ define_int CONFIG_STACK_SIZE_SHIFT 3 ++ else ++ if [ "$CONFIG_STACK_SIZE_64KB" = "y" ]; then ++ define_int CONFIG_STACK_SIZE_SHIFT 4 ++ fi ++ fi ++ fi ++fi ++ + endmenu + + mainmenu_option next_comment +Index: linux-2.4.21-171/arch/i386/vmlinux.lds.S +=================================================================== +--- linux-2.4.21-171.orig/arch/i386/vmlinux.lds.S 2004-02-24 13:43:40.000000000 -0500 ++++ linux-2.4.21-171/arch/i386/vmlinux.lds.S 2004-04-03 16:02:32.000000000 -0500 +@@ -39,7 +39,8 @@ + + _edata = .; /* End of data section */ + +- . = ALIGN(8192); /* init_task */ ++/* chose the biggest of the possible stack sizes here? */ ++ . = ALIGN(65536); /* init_task */ + .data.init_task : { *(.data.init_task) } + + . = ALIGN(4096); /* Init code and data */ +Index: linux-2.4.21-171/include/asm-i386/current.h +=================================================================== +--- linux-2.4.21-171.orig/include/asm-i386/current.h 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/include/asm-i386/current.h 2004-04-03 16:02:32.000000000 -0500 +@@ -1,15 +1,43 @@ + #ifndef _I386_CURRENT_H + #define _I386_CURRENT_H ++#include ++ ++/* ++ * Configurable page sizes on i386, mainly for debugging purposes. ++ * (c) Balbir Singh ++ */ ++ ++#ifdef __ASSEMBLY__ ++ ++#define PAGE_SIZE 4096 /* as cannot handle 1UL << 12 */ ++#define THREAD_SIZE ((1 << CONFIG_STACK_SIZE_SHIFT) * PAGE_SIZE) ++ ++#define GET_CURRENT(reg) \ ++ movl $-THREAD_SIZE, reg; \ ++ andl %esp, reg ++ ++#else /* __ASSEMBLY__ */ ++ ++#define THREAD_SIZE ((1 << CONFIG_STACK_SIZE_SHIFT) * PAGE_SIZE) ++#define alloc_task_struct() \ ++ ((struct task_struct *) __get_free_pages(GFP_KERNEL,CONFIG_STACK_SIZE_SHIFT)) ++ ++#define free_task_struct(p) \ ++ free_pages((unsigned long) (p), CONFIG_STACK_SIZE_SHIFT) ++ ++#define INIT_TASK_SIZE THREAD_SIZE + + struct task_struct; + + static inline struct task_struct * get_current(void) + { + struct task_struct *current; +- __asm__("andl %%esp,%0; ":"=r" (current) : "0" (~8191UL)); ++ __asm__("andl %%esp,%0; ":"=r" (current) : "0" (~(THREAD_SIZE - 1))); + return current; + } + + #define current get_current() + ++#endif /* __ASSEMBLY__ */ ++ + #endif /* !(_I386_CURRENT_H) */ +Index: linux-2.4.21-171/include/asm-i386/hw_irq.h +=================================================================== +--- linux-2.4.21-171.orig/include/asm-i386/hw_irq.h 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/include/asm-i386/hw_irq.h 2004-04-03 16:02:32.000000000 -0500 +@@ -16,6 +16,7 @@ + #include + #include + #include ++#include + + /* + * IDT vectors usable for external interrupt sources start +@@ -116,10 +117,6 @@ + #define IRQ_NAME2(nr) nr##_interrupt(void) + #define IRQ_NAME(nr) IRQ_NAME2(IRQ##nr) + +-#define GET_CURRENT \ +- "movl %esp, %ebx\n\t" \ +- "andl $-8192, %ebx\n\t" +- + /* + * SMP has a few special interrupts for IPI messages + */ +Index: linux-2.4.21-171/include/asm-i386/processor.h +=================================================================== +--- linux-2.4.21-171.orig/include/asm-i386/processor.h 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/include/asm-i386/processor.h 2004-04-03 16:02:32.000000000 -0500 +@@ -14,6 +14,7 @@ + #include + #include + #include ++#include + #include + #include + #include +@@ -464,9 +465,6 @@ + #define KSTK_EIP(tsk) (((unsigned long *)(4096+(unsigned long)(tsk)))[1019]) + #define KSTK_ESP(tsk) (((unsigned long *)(4096+(unsigned long)(tsk)))[1022]) + +-#define THREAD_SIZE (2*PAGE_SIZE) +-#define alloc_task_struct() ((struct task_struct *) __get_free_pages(GFP_KERNEL,1)) +-#define free_task_struct(p) free_pages((unsigned long) (p), 1) + #define get_task_struct(tsk) atomic_inc(&virt_to_page(tsk)->count) + + #define init_task (init_task_union.task) +Index: linux-2.4.21-171/include/linux/sched.h +=================================================================== +--- linux-2.4.21-171.orig/include/linux/sched.h 2004-03-31 14:58:26.000000000 -0500 ++++ linux-2.4.21-171/include/linux/sched.h 2004-04-03 16:02:32.000000000 -0500 +@@ -2,6 +2,7 @@ + #define _LINUX_SCHED_H + + #include /* for HZ */ ++#include /* maybe for INIT_TASK_SIZE */ + + extern unsigned long event; + +Index: linux-2.4.21-171/include/asm-x86_64/current.h +=================================================================== +--- linux-2.4.21-171.orig/include/asm-x86_64/current.h 2004-03-31 14:58:26.000000000 -0500 ++++ linux-2.4.21-171/include/asm-x86_64/current.h 2004-04-03 16:02:32.000000000 -0500 +@@ -5,6 +5,7 @@ + struct task_struct; + + #include ++#include + + static inline struct task_struct *get_current(void) + { diff --git a/lustre/kernel_patches/patches/tcp-zero-copy-2.4.21-suse-171.patch b/lustre/kernel_patches/patches/tcp-zero-copy-2.4.21-suse-171.patch new file mode 100644 index 0000000..8ef6cd2 --- /dev/null +++ b/lustre/kernel_patches/patches/tcp-zero-copy-2.4.21-suse-171.patch @@ -0,0 +1,456 @@ +Index: linux-2.4.21-171/include/linux/skbuff.h +=================================================================== +--- linux-2.4.21-171.orig/include/linux/skbuff.h 2004-03-31 14:58:38.000000000 -0500 ++++ linux-2.4.21-171/include/linux/skbuff.h 2004-04-03 16:43:15.000000000 -0500 +@@ -116,6 +116,30 @@ + __u16 size; + }; + ++/* Support for callback when skb data has been released */ ++typedef struct zccd /* Zero Copy Callback Descriptor */ ++{ /* (embed as first member of custom struct) */ ++ atomic_t zccd_count; /* reference count */ ++ void (*zccd_destructor)(struct zccd *); /* callback when refcount reaches zero */ ++} zccd_t; ++ ++static inline void zccd_init (zccd_t *d, void (*callback)(zccd_t *)) ++{ ++ atomic_set (&d->zccd_count, 1); ++ d->zccd_destructor = callback; ++} ++ ++static inline void zccd_get (zccd_t *d) /* take a reference */ ++{ ++ atomic_inc (&d->zccd_count); ++} ++ ++static inline void zccd_put (zccd_t *d) /* release a reference */ ++{ ++ if (atomic_dec_and_test (&d->zccd_count)) ++ (d->zccd_destructor)(d); ++} ++ + /* This data is invariant across clones and lives at + * the end of the header data, ie. at skb->end. + */ +@@ -123,6 +147,12 @@ + atomic_t dataref; + unsigned int nr_frags; + struct sk_buff *frag_list; ++ zccd_t *zccd; /* zero copy descriptor */ ++ zccd_t *zccd2; /* 2nd zero copy descriptor */ ++ /* NB we expect zero-copy data to be at least 1 packet, so ++ * having 2 zccds means we don't unneccessarily split the packet ++ * where consecutive zero-copy sends abutt. ++ */ + skb_frag_t frags[MAX_SKB_FRAGS]; + }; + +Index: linux-2.4.21-171/include/net/tcp.h +=================================================================== +--- linux-2.4.21-171.orig/include/net/tcp.h 2004-03-31 15:07:31.000000000 -0500 ++++ linux-2.4.21-171/include/net/tcp.h 2004-04-03 16:46:15.000000000 -0500 +@@ -646,6 +646,8 @@ + + extern int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size); + extern ssize_t tcp_sendpage(struct socket *sock, struct page *page, int offset, size_t size, int flags); ++extern ssize_t tcp_sendpage_zccd(struct socket *sock, struct page *page, int offset, ssize_t size, ++ int flags, zccd_t *zccd); + + extern int tcp_ioctl(struct sock *sk, + int cmd, +@@ -742,6 +744,9 @@ + struct msghdr *msg, + int len, int nonblock, + int flags, int *addr_len); ++extern int tcp_recvpackets(struct sock *sk, ++ struct sk_buff_head *packets, ++ int len, int nonblock); + + extern int tcp_listen_start(struct sock *sk); + +Index: linux-2.4.21-171/net/netsyms.c +=================================================================== +--- linux-2.4.21-171.orig/net/netsyms.c 2004-02-24 14:03:22.000000000 -0500 ++++ linux-2.4.21-171/net/netsyms.c 2004-04-03 16:13:53.000000000 -0500 +@@ -407,6 +407,8 @@ + + #endif + ++EXPORT_SYMBOL(tcp_sendpage_zccd); ++EXPORT_SYMBOL(tcp_recvpackets); + EXPORT_SYMBOL(tcp_read_sock); + + EXPORT_SYMBOL(netlink_set_err); +Index: linux-2.4.21-171/net/core/skbuff.c +=================================================================== +--- linux-2.4.21-171.orig/net/core/skbuff.c 2004-02-24 14:03:22.000000000 -0500 ++++ linux-2.4.21-171/net/core/skbuff.c 2004-04-03 16:13:53.000000000 -0500 +@@ -208,6 +208,8 @@ + atomic_set(&(skb_shinfo(skb)->dataref), 1); + skb_shinfo(skb)->nr_frags = 0; + skb_shinfo(skb)->frag_list = NULL; ++ skb_shinfo(skb)->zccd = NULL; /* skbuffs kick off with NO user zero copy descriptors */ ++ skb_shinfo(skb)->zccd2 = NULL; + return skb; + + nodata: +@@ -277,6 +279,10 @@ + { + if (!skb->cloned || + atomic_dec_and_test(&(skb_shinfo(skb)->dataref))) { ++ if (skb_shinfo(skb)->zccd != NULL) /* zero copy callback descriptor? */ ++ zccd_put (skb_shinfo(skb)->zccd); /* release hold */ ++ if (skb_shinfo(skb)->zccd2 != NULL) /* 2nd zero copy callback descriptor? */ ++ zccd_put (skb_shinfo(skb)->zccd2); /* release hold */ + if (skb_shinfo(skb)->nr_frags) { + int i; + for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) +@@ -535,6 +541,8 @@ + atomic_set(&(skb_shinfo(skb)->dataref), 1); + skb_shinfo(skb)->nr_frags = 0; + skb_shinfo(skb)->frag_list = NULL; ++ skb_shinfo(skb)->zccd = NULL; /* copied data => no user zero copy descriptor */ ++ skb_shinfo(skb)->zccd2 = NULL; + + /* We are no longer a clone, even if we were. */ + skb->cloned = 0; +@@ -581,6 +589,14 @@ + n->data_len = skb->data_len; + n->len = skb->len; + ++ if (skb_shinfo(skb)->zccd != NULL) /* user zero copy descriptor? */ ++ zccd_get (skb_shinfo(skb)->zccd); /* 1 more ref (pages are shared) */ ++ skb_shinfo(n)->zccd = skb_shinfo(skb)->zccd; ++ ++ if (skb_shinfo(skb)->zccd2 != NULL) /* 2nd user zero copy descriptor? */ ++ zccd_get (skb_shinfo(skb)->zccd2); /* 1 more ref (pages are shared) */ ++ skb_shinfo(n)->zccd2 = skb_shinfo(skb)->zccd2; ++ + if (skb_shinfo(skb)->nr_frags) { + int i; + +@@ -623,6 +639,8 @@ + u8 *data; + int size = nhead + (skb->end - skb->head) + ntail; + long off; ++ zccd_t *zccd = skb_shinfo(skb)->zccd; /* stash user zero copy descriptor */ ++ zccd_t *zccd2 = skb_shinfo(skb)->zccd2; /* stash 2nd user zero copy descriptor */ + + if (skb_shared(skb)) + BUG(); +@@ -644,6 +662,11 @@ + if (skb_shinfo(skb)->frag_list) + skb_clone_fraglist(skb); + ++ if (zccd != NULL) /* user zero copy descriptor? */ ++ zccd_get (zccd); /* extra ref (pages are shared) */ ++ if (zccd2 != NULL) /* 2nd user zero copy descriptor? */ ++ zccd_get (zccd2); /* extra ref (pages are shared) */ ++ + skb_release_data(skb); + + off = (data+nhead) - skb->head; +@@ -658,6 +681,8 @@ + skb->nh.raw += off; + skb->cloned = 0; + atomic_set(&skb_shinfo(skb)->dataref, 1); ++ skb_shinfo(skb)->zccd = zccd; ++ skb_shinfo(skb)->zccd2 = zccd2; + return 0; + + nodata: +Index: linux-2.4.21-171/net/ipv4/tcp.c +=================================================================== +--- linux-2.4.21-171.orig/net/ipv4/tcp.c 2004-02-24 13:42:30.000000000 -0500 ++++ linux-2.4.21-171/net/ipv4/tcp.c 2004-04-03 16:43:05.000000000 -0500 +@@ -748,7 +748,7 @@ + goto out; + } + +-ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size_t psize, int flags); ++ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size_t psize, int flags, zccd_t *zccd); + + static inline int + can_coalesce(struct sk_buff *skb, int i, struct page *page, int off) +@@ -827,7 +827,7 @@ + return err; + } + +-ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size_t psize, int flags) ++ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size_t psize, int flags, zccd_t *zccd) + { + struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp); + int mss_now; +@@ -875,6 +875,17 @@ + copy = size; + + i = skb_shinfo(skb)->nr_frags; ++ ++ if (zccd != NULL && /* this is a zcc I/O */ ++ skb_shinfo(skb)->zccd != NULL && /* skb is part of a zcc I/O */ ++ skb_shinfo(skb)->zccd2 != NULL && ++ skb_shinfo(skb)->zccd != zccd && /* not the same one */ ++ skb_shinfo(skb)->zccd2 != zccd) ++ { ++ tcp_mark_push (tp, skb); ++ goto new_segment; ++ } ++ + if (can_coalesce(skb, i, page, offset)) { + skb_shinfo(skb)->frags[i-1].size += copy; + } else if (i < MAX_SKB_FRAGS) { +@@ -885,6 +896,20 @@ + goto new_segment; + } + ++ if (zccd != NULL && /* this is a zcc I/O */ ++ skb_shinfo(skb)->zccd != zccd && /* not already referencing this zccd */ ++ skb_shinfo(skb)->zccd2 != zccd) ++ { ++ zccd_get (zccd); /* bump ref count */ ++ ++ BUG_TRAP (skb_shinfo(skb)->zccd2 == NULL); ++ ++ if (skb_shinfo(skb)->zccd == NULL) /* reference this zccd */ ++ skb_shinfo(skb)->zccd = zccd; ++ else ++ skb_shinfo(skb)->zccd2 = zccd; ++ } ++ + skb->len += copy; + skb->data_len += copy; + skb->ip_summed = CHECKSUM_HW; +@@ -948,7 +973,28 @@ + + lock_sock(sk); + TCP_CHECK_TIMER(sk); +- res = do_tcp_sendpages(sk, &page, offset, size, flags); ++ res = do_tcp_sendpages(sk, &page, offset, size, flags, NULL); ++ TCP_CHECK_TIMER(sk); ++ release_sock(sk); ++ return res; ++} ++ ++ssize_t tcp_sendpage_zccd(struct socket *sock, struct page *page, int offset, ssize_t size, ++ int flags, zccd_t *zccd) ++{ ++ ssize_t res; ++ struct sock *sk = sock->sk; ++ ++#define TCP_ZC_CSUM_FLAGS (NETIF_F_IP_CSUM|NETIF_F_NO_CSUM|NETIF_F_HW_CSUM) ++ ++ if (!(sk->route_caps & NETIF_F_SG) || /* caller shouldn't waste her time */ ++ !(sk->route_caps & TCP_ZC_CSUM_FLAGS)) /* on double mapping */ ++ BUG (); ++ ++ lock_sock(sk); ++ TCP_CHECK_TIMER(sk); ++ ++ res = do_tcp_sendpages(sk, &page, 0, size, flags, zccd); + TCP_CHECK_TIMER(sk); + release_sock(sk); + return res; +@@ -1772,6 +1818,202 @@ + goto out; + } + ++int tcp_recvpackets (struct sock *sk, struct sk_buff_head *packets, ++ int len, int nonblock) ++{ ++ struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp); ++ int copied; ++ long timeo; ++ ++ BUG_TRAP (len > 0); ++ /*BUG_TRAP ((flags & (MSG_OOB | MSG_PEEK | MSG_TRUNC)) == 0);*/ ++ ++ lock_sock(sk); ++ ++ TCP_CHECK_TIMER(sk); ++ ++ copied = -ENOTCONN; ++ if (sk->state == TCP_LISTEN) ++ goto out; ++ ++ copied = 0; ++ timeo = sock_rcvtimeo(sk, nonblock); ++ ++ do { ++ struct sk_buff * skb; ++ u32 offset; ++ unsigned long used; ++ int exhausted; ++ int eaten; ++ ++ /* Are we at urgent data? Stop if we have read anything. */ ++ if (copied && tp->urg_data && tp->urg_seq == tp->copied_seq) ++ break; ++ ++ /* We need to check signals first, to get correct SIGURG ++ * handling. FIXME: Need to check this doesnt impact 1003.1g ++ * and move it down to the bottom of the loop ++ */ ++ if (signal_pending(current)) { ++ if (copied) ++ break; ++ copied = timeo ? sock_intr_errno(timeo) : -EAGAIN; ++ break; ++ } ++ ++ /* Next get a buffer. */ ++ ++ skb = skb_peek(&sk->receive_queue); ++ ++ if (skb == NULL) /* nothing ready */ ++ { ++ if (copied) { ++ if (sk->err || ++ sk->state == TCP_CLOSE || ++ (sk->shutdown & RCV_SHUTDOWN) || ++ !timeo || ++ (0)) ++ break; ++ } else { ++ if (sk->done) ++ break; ++ ++ if (sk->err) { ++ copied = sock_error(sk); ++ break; ++ } ++ ++ if (sk->shutdown & RCV_SHUTDOWN) ++ break; ++ ++ if (sk->state == TCP_CLOSE) { ++ if (!sk->done) { ++ /* This occurs when user tries to read ++ * from never connected socket. ++ */ ++ copied = -ENOTCONN; ++ break; ++ } ++ break; ++ } ++ ++ if (!timeo) { ++ copied = -EAGAIN; ++ break; ++ } ++ } ++ ++ cleanup_rbuf(sk, copied); ++ timeo = tcp_data_wait(sk, timeo); ++ continue; ++ } ++ ++ BUG_TRAP (atomic_read (&skb->users) == 1); ++ ++ exhausted = eaten = 0; ++ ++ offset = tp->copied_seq - TCP_SKB_CB(skb)->seq; ++ if (skb->h.th->syn) ++ offset--; ++ ++ used = skb->len - offset; ++ ++ if (tp->urg_data) { ++ u32 urg_offset = tp->urg_seq - tp->copied_seq; ++ if (urg_offset < used) { ++ if (!urg_offset) { /* at urgent date */ ++ if (!sk->urginline) { ++ tp->copied_seq++; /* discard the single byte of urgent data */ ++ offset++; ++ used--; ++ } ++ } else /* truncate read */ ++ used = urg_offset; ++ } ++ } ++ ++ BUG_TRAP (used >= 0); ++ if (len < used) ++ used = len; ++ ++ if (used == 0) ++ exhausted = 1; ++ else ++ { ++ if (skb_is_nonlinear (skb)) ++ { ++ int rc = skb_linearize (skb, GFP_KERNEL); ++ ++ printk ("tcp_recvpackets(): linearising: %d\n", rc); ++ ++ if (rc) ++ { ++ if (!copied) ++ copied = rc; ++ break; ++ } ++ } ++ ++ if ((offset + used) == skb->len) /* consuming the whole packet */ ++ { ++ __skb_unlink (skb, &sk->receive_queue); ++ dst_release (skb->dst); ++ skb_orphan (skb); ++ __skb_pull (skb, offset); ++ __skb_queue_tail (packets, skb); ++ exhausted = eaten = 1; ++ } ++ else /* consuming only part of the packet */ ++ { ++ struct sk_buff *skb2 = skb_clone (skb, GFP_KERNEL); ++ ++ if (skb2 == NULL) ++ { ++ if (!copied) ++ copied = -ENOMEM; ++ break; ++ } ++ ++ dst_release (skb2->dst); ++ __skb_pull (skb2, offset); ++ __skb_trim (skb2, used); ++ __skb_queue_tail (packets, skb2); ++ } ++ ++ tp->copied_seq += used; ++ copied += used; ++ len -= used; ++ } ++ ++ if (tp->urg_data && after(tp->copied_seq,tp->urg_seq)) { ++ tp->urg_data = 0; ++ tcp_fast_path_check(sk, tp); ++ } ++ ++ if (!exhausted) ++ continue; ++ ++ if (skb->h.th->fin) ++ { ++ tp->copied_seq++; ++ if (!eaten) ++ tcp_eat_skb (sk, skb); ++ break; ++ } ++ ++ if (!eaten) ++ tcp_eat_skb (sk, skb); ++ ++ } while (len > 0); ++ ++ out: ++ /* Clean up data we have read: This will do ACK frames. */ ++ cleanup_rbuf(sk, copied); ++ TCP_CHECK_TIMER(sk); ++ release_sock(sk); ++ return copied; ++} ++ + /* + * State processing on a close. This implements the state shift for + * sending our FIN frame. Note that we only send a FIN for some diff --git a/lustre/kernel_patches/patches/vfs_intent-2.4.21-suse-171.patch b/lustre/kernel_patches/patches/vfs_intent-2.4.21-suse-171.patch new file mode 100644 index 0000000..754e756 --- /dev/null +++ b/lustre/kernel_patches/patches/vfs_intent-2.4.21-suse-171.patch @@ -0,0 +1,1847 @@ + fs/dcache.c | 19 ++ + fs/exec.c | 17 +- + fs/namei.c | 295 +++++++++++++++++++++++++++++++++++++++------- + fs/namespace.c | 28 +++- + fs/open.c | 172 +++++++++++++++++++------- + fs/stat.c | 52 +++++--- + include/linux/dcache.h | 60 +++++++++ + include/linux/fs.h | 32 ++++ + include/linux/fs_struct.h | 4 + kernel/exit.c | 3 + kernel/fork.c | 3 + kernel/ksyms.c | 1 + 12 files changed, 558 insertions(+), 128 deletions(-) + +Index: linux-2.4.21-171/fs/dcache.c +=================================================================== +--- linux-2.4.21-171.orig/fs/dcache.c 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/fs/dcache.c 2004-04-03 16:09:50.000000000 -0500 +@@ -186,6 +186,13 @@ + spin_unlock(&dcache_lock); + return 0; + } ++ ++ /* network invalidation by Lustre */ ++ if (dentry->d_flags & DCACHE_LUSTRE_INVALID) { ++ spin_unlock(&dcache_lock); ++ return 0; ++ } ++ + /* + * Check whether to do a partial shrink_dcache + * to get rid of unused child entries. +@@ -838,13 +845,19 @@ + * Adds a dentry to the hash according to its name. + */ + +-void d_rehash(struct dentry * entry) ++void __d_rehash(struct dentry * entry, int lock) + { + struct list_head *list = d_hash(entry->d_parent, entry->d_name.hash); + if (!list_empty(&entry->d_hash)) BUG(); +- spin_lock(&dcache_lock); ++ if (lock) spin_lock(&dcache_lock); + list_add(&entry->d_hash, list); +- spin_unlock(&dcache_lock); ++ if (lock) spin_unlock(&dcache_lock); ++} ++EXPORT_SYMBOL(__d_rehash); ++ ++void d_rehash(struct dentry * entry) ++{ ++ __d_rehash(entry, 1); + } + + #define do_switch(x,y) do { \ +Index: linux-2.4.21-171/fs/exec.c +=================================================================== +--- linux-2.4.21-171.orig/fs/exec.c 2004-02-24 14:03:21.000000000 -0500 ++++ linux-2.4.21-171/fs/exec.c 2004-04-03 16:09:50.000000000 -0500 +@@ -113,8 +113,10 @@ + struct file * file; + struct nameidata nd; + int error; ++ struct lookup_intent it = { .it_op = IT_OPEN, ++ .it_flags = FMODE_READ|FMODE_EXEC }; + +- error = user_path_walk(library, &nd); ++ error = user_path_walk_it(library, &nd, &it); + if (error) + goto out; + +@@ -126,7 +128,8 @@ + if (error) + goto exit; + +- file = dentry_open(nd.dentry, nd.mnt, O_RDONLY); ++ file = dentry_open_it(nd.dentry, nd.mnt, O_RDONLY, &it); ++ intent_release(&it); + error = PTR_ERR(file); + if (IS_ERR(file)) + goto out; +@@ -383,8 +386,10 @@ + struct inode *inode; + struct file *file; + int err = 0; ++ struct lookup_intent it = { .it_op = IT_OPEN, ++ .it_flags = FMODE_READ|FMODE_EXEC }; + +- err = path_lookup(name, LOOKUP_FOLLOW|LOOKUP_POSITIVE, &nd); ++ err = path_lookup_it(name, LOOKUP_FOLLOW|LOOKUP_POSITIVE, &nd, &it); + file = ERR_PTR(err); + if (!err) { + inode = nd.dentry->d_inode; +@@ -396,7 +401,8 @@ + err = -EACCES; + file = ERR_PTR(err); + if (!err) { +- file = dentry_open(nd.dentry, nd.mnt, O_RDONLY); ++ file = dentry_open_it(nd.dentry, nd.mnt, O_RDONLY, &it); ++ intent_release(&it); + if (!IS_ERR(file)) { + err = deny_write_access(file); + if (err) { +@@ -408,6 +414,7 @@ + return file; + } + } ++ intent_release(&it); + path_release(&nd); + } + goto out; +@@ -1148,7 +1155,7 @@ + goto close_fail; + if (!file->f_op->write) + goto close_fail; +- if (do_truncate(file->f_dentry, 0) != 0) ++ if (do_truncate(file->f_dentry, 0, 0) != 0) + goto close_fail; + + retval = binfmt->core_dump(signr, regs, file); +Index: linux-2.4.21-171/fs/namei.c +=================================================================== +--- linux-2.4.21-171.orig/fs/namei.c 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/fs/namei.c 2004-04-03 16:13:23.000000000 -0500 +@@ -94,6 +94,13 @@ + * XEmacs seems to be relying on it... + */ + ++void intent_release(struct lookup_intent *it) ++{ ++ if (it && it->it_op_release) ++ it->it_op_release(it); ++ ++} ++ + /* In order to reduce some races, while at the same time doing additional + * checking and hopefully speeding things up, we copy filenames to the + * kernel data space before using them.. +@@ -260,10 +267,19 @@ + * Internal lookup() using the new generic dcache. + * SMP-safe + */ +-static struct dentry * cached_lookup(struct dentry * parent, struct qstr * name, int flags) ++static struct dentry *cached_lookup(struct dentry *parent, struct qstr *name, ++ int flags, struct lookup_intent *it) + { + struct dentry * dentry = d_lookup(parent, name); + ++ if (dentry && dentry->d_op && dentry->d_op->d_revalidate_it) { ++ if (!dentry->d_op->d_revalidate_it(dentry, flags, it) && ++ !d_invalidate(dentry)) { ++ dput(dentry); ++ dentry = NULL; ++ } ++ return dentry; ++ } else + if (dentry && dentry->d_op && dentry->d_op->d_revalidate) { + if (!dentry->d_op->d_revalidate(dentry, flags) && !d_invalidate(dentry)) { + dput(dentry); +@@ -281,11 +297,15 @@ + * make sure that nobody added the entry to the dcache in the meantime.. + * SMP-safe + */ +-static struct dentry * real_lookup(struct dentry * parent, struct qstr * name, int flags) ++static struct dentry *real_lookup(struct dentry *parent, struct qstr *name, ++ int flags, struct lookup_intent *it) + { + struct dentry * result; + struct inode *dir = parent->d_inode; ++ int counter = 0; + ++again: ++ counter++; + down(&dir->i_sem); + /* + * First re-do the cached lookup just in case it was created +@@ -300,6 +320,9 @@ + result = ERR_PTR(-ENOMEM); + if (dentry) { + lock_kernel(); ++ if (dir->i_op->lookup_it) ++ result = dir->i_op->lookup_it(dir, dentry, it, flags); ++ else + result = dir->i_op->lookup(dir, dentry); + unlock_kernel(); + if (result) +@@ -321,6 +344,15 @@ + dput(result); + result = ERR_PTR(-ENOENT); + } ++ } else if (result->d_op && result->d_op->d_revalidate_it) { ++ if (!result->d_op->d_revalidate_it(result, flags, it) && ++ !d_invalidate(result)) { ++ dput(result); ++ if (counter > 10) ++ result = ERR_PTR(-ESTALE); ++ if (!IS_ERR(result)) ++ goto again; ++ } + } + return result; + } +@@ -332,7 +364,8 @@ + * Without that kind of total limit, nasty chains of consecutive + * symlinks can cause almost arbitrarily long lookups. + */ +-static inline int do_follow_link(struct dentry *dentry, struct nameidata *nd) ++static inline int do_follow_link(struct dentry *dentry, struct nameidata *nd, ++ struct lookup_intent *it) + { + int err; + if (current->link_count >= 8) +@@ -346,10 +379,12 @@ + current->link_count++; + current->total_link_count++; + UPDATE_ATIME(dentry->d_inode); ++ nd->intent = it; + err = dentry->d_inode->i_op->follow_link(dentry, nd); + current->link_count--; + return err; + loop: ++ intent_release(it); + path_release(nd); + return -ELOOP; + } +@@ -448,7 +483,8 @@ + * We expect 'base' to be positive and a directory. + */ + static inline int __attribute__((always_inline)) +-__link_path_walk(const char * name, struct nameidata *nd) ++__link_path_walk_it(const char * name, struct nameidata *nd, ++ struct lookup_intent *it) + { + struct dentry *dentry; + struct inode *inode; +@@ -525,12 +561,12 @@ + break; + } + /* This does the actual lookups.. */ +- dentry = cached_lookup(nd->dentry, &this, LOOKUP_CONTINUE); ++ dentry = cached_lookup(nd->dentry, &this, LOOKUP_CONTINUE, NULL); + if (!dentry) { + err = -EWOULDBLOCKIO; + if (atomic) + break; +- dentry = real_lookup(nd->dentry, &this, LOOKUP_CONTINUE); ++ dentry = real_lookup(nd->dentry, &this, LOOKUP_CONTINUE, NULL); + err = PTR_ERR(dentry); + if (IS_ERR(dentry)) + break; +@@ -548,7 +584,7 @@ + goto out_dput; + + if (inode->i_op->follow_link) { +- err = do_follow_link(dentry, nd); ++ err = do_follow_link(dentry, nd, NULL); + dput(dentry); + if (err) + goto return_err; +@@ -564,7 +600,7 @@ + nd->dentry = dentry; + } + err = -ENOTDIR; +- if (!inode->i_op->lookup) ++ if (!inode->i_op->lookup && !inode->i_op->lookup_it) + break; + continue; + /* here ends the main loop */ +@@ -591,12 +627,12 @@ + if (err < 0) + break; + } +- dentry = cached_lookup(nd->dentry, &this, 0); ++ dentry = cached_lookup(nd->dentry, &this, 0, it); + if (!dentry) { + err = -EWOULDBLOCKIO; + if (atomic) + break; +- dentry = real_lookup(nd->dentry, &this, 0); ++ dentry = real_lookup(nd->dentry, &this, 0, it); + err = PTR_ERR(dentry); + if (IS_ERR(dentry)) + break; +@@ -606,7 +642,7 @@ + inode = dentry->d_inode; + if ((lookup_flags & LOOKUP_FOLLOW) + && inode && inode->i_op && inode->i_op->follow_link) { +- err = do_follow_link(dentry, nd); ++ err = do_follow_link(dentry, nd, it); + dput(dentry); + if (err) + goto return_err; +@@ -620,7 +656,8 @@ + goto no_inode; + if (lookup_flags & LOOKUP_DIRECTORY) { + err = -ENOTDIR; +- if (!inode->i_op || !inode->i_op->lookup) ++ if (!inode->i_op || ++ (!inode->i_op->lookup && !inode->i_op->lookup_it)) + break; + } + goto return_base; +@@ -644,6 +681,25 @@ + * Check the cached dentry for staleness. + */ + dentry = nd->dentry; ++ if (dentry && dentry->d_op && dentry->d_op->d_revalidate_it) { ++ err = -ESTALE; ++ if (!dentry->d_op->d_revalidate_it(dentry, 0, it)) { ++ struct dentry *new; ++ err = permission(dentry->d_parent->d_inode, ++ MAY_EXEC); ++ if (err) ++ break; ++ new = real_lookup(dentry->d_parent, ++ &dentry->d_name, 0, NULL); ++ d_invalidate(dentry); ++ dput(dentry); ++ if (IS_ERR(new)) { ++ err = PTR_ERR(new); ++ break; ++ } ++ nd->dentry = new; ++ } ++ } else + if (dentry && dentry->d_op && dentry->d_op->d_revalidate) { + err = -ESTALE; + if (!dentry->d_op->d_revalidate(dentry, lookup_flags & LOOKUP_PARENT)) { +@@ -657,6 +713,8 @@ + dput(dentry); + break; + } ++ if (err) ++ intent_release(it); + path_release(nd); + return_err: + return err; +@@ -664,13 +722,13 @@ + + int link_path_walk(const char * name, struct nameidata *nd) + { +- return __link_path_walk(name,nd); ++ return __link_path_walk_it(name, nd, NULL); + } + + static inline int __path_walk(const char * name, struct nameidata *nd) + { + current->total_link_count = 0; +- return __link_path_walk(name, nd); ++ return __link_path_walk_it(name, nd, NULL); + } + + int path_walk(const char * name, struct nameidata *nd) +@@ -678,6 +736,12 @@ + return __path_walk(name, nd); + } + ++int path_walk_it(const char * name, struct nameidata *nd, struct lookup_intent *it) ++{ ++ current->total_link_count = 0; ++ return __link_path_walk_it(name, nd, it); ++} ++ + /* SMP-safe */ + /* returns 1 if everything is done */ + static int __emul_lookup_dentry(const char *name, struct nameidata *nd) +@@ -760,6 +824,17 @@ + } + + /* SMP-safe */ ++int path_lookup_it(const char *path, unsigned flags, struct nameidata *nd, ++ struct lookup_intent *it) ++{ ++ int error = 0; ++ if (path_init(path, flags, nd)) ++ error = path_walk_it(path, nd, it); ++ return error; ++} ++ ++ ++/* SMP-safe */ + int path_lookup(const char *path, unsigned flags, struct nameidata *nd) + { + int error = 0; +@@ -774,6 +849,7 @@ + { + nd->last_type = LAST_ROOT; /* if there are only slashes... */ + nd->flags = flags; ++ nd->intent = NULL; + if (*name=='/') + return walk_init_root(name,nd); + read_lock(¤t->fs->lock); +@@ -788,7 +864,8 @@ + * needs parent already locked. Doesn't follow mounts. + * SMP-safe. + */ +-struct dentry * lookup_hash(struct qstr *name, struct dentry * base) ++struct dentry * lookup_hash_it(struct qstr *name, struct dentry * base, ++ struct lookup_intent *it) + { + struct dentry * dentry; + struct inode *inode; +@@ -811,13 +888,16 @@ + goto out; + } + +- dentry = cached_lookup(base, name, 0); ++ dentry = cached_lookup(base, name, 0, it); + if (!dentry) { + struct dentry *new = d_alloc(base, name); + dentry = ERR_PTR(-ENOMEM); + if (!new) + goto out; + lock_kernel(); ++ if (inode->i_op->lookup_it) ++ dentry = inode->i_op->lookup_it(inode, new, it, 0); ++ else + dentry = inode->i_op->lookup(inode, new); + unlock_kernel(); + if (!dentry) +@@ -829,6 +909,12 @@ + return dentry; + } + ++struct dentry * lookup_hash(struct qstr *name, struct dentry * base) ++{ ++ return lookup_hash_it(name, base, NULL); ++} ++ ++ + /* SMP-safe */ + struct dentry * lookup_one_len(const char * name, struct dentry * base, int len) + { +@@ -850,7 +936,7 @@ + } + this.hash = end_name_hash(hash); + +- return lookup_hash(&this, base); ++ return lookup_hash_it(&this, base, NULL); + access: + return ERR_PTR(-EACCES); + } +@@ -881,6 +967,23 @@ + return err; + } + ++int __user_walk_it(const char *name, unsigned flags, struct nameidata *nd, ++ struct lookup_intent *it) ++{ ++ char *tmp; ++ int err; ++ ++ tmp = getname(name); ++ err = PTR_ERR(tmp); ++ if (!IS_ERR(tmp)) { ++ err = 0; ++ if (path_init(tmp, flags, nd)) ++ err = path_walk_it(tmp, nd, it); ++ putname(tmp); ++ } ++ return err; ++} ++ + /* + * It's inline, so penalty for filesystems that don't use sticky bit is + * minimal. +@@ -978,7 +1081,8 @@ + return retval; + } + +-int vfs_create(struct inode *dir, struct dentry *dentry, int mode) ++static int vfs_create_it(struct inode *dir, struct dentry *dentry, int mode, ++ struct lookup_intent *it) + { + int error; + +@@ -991,12 +1095,15 @@ + goto exit_lock; + + error = -EACCES; /* shouldn't it be ENOSYS? */ +- if (!dir->i_op || !dir->i_op->create) ++ if (!dir->i_op || (!dir->i_op->create && !dir->i_op->create_it)) + goto exit_lock; + + DQUOT_INIT(dir); + lock_kernel(); +- error = dir->i_op->create(dir, dentry, mode); ++ if (dir->i_op->create_it) ++ error = dir->i_op->create_it(dir, dentry, mode, it); ++ else ++ error = dir->i_op->create(dir, dentry, mode); + unlock_kernel(); + exit_lock: + up(&dir->i_zombie); +@@ -1005,6 +1112,11 @@ + return error; + } + ++int vfs_create(struct inode *dir, struct dentry *dentry, int mode) ++{ ++ return vfs_create_it(dir, dentry, mode, NULL); ++} ++ + /* + * open_namei() + * +@@ -1019,7 +1131,8 @@ + * for symlinks (where the permissions are checked later). + * SMP-safe + */ +-int open_namei(const char * pathname, int flag, int mode, struct nameidata *nd) ++int open_namei_it(const char *pathname, int flag, int mode, ++ struct nameidata *nd, struct lookup_intent *it) + { + int acc_mode, error = 0; + struct inode *inode; +@@ -1029,11 +1142,14 @@ + + acc_mode = ACC_MODE(flag); + ++ if (it) ++ it->it_flags = flag; ++ + /* + * The simplest case - just a plain lookup. + */ + if (!(flag & O_CREAT)) { +- error = path_lookup(pathname, lookup_flags(flag), nd); ++ error = path_lookup_it(pathname, lookup_flags(flag), nd, it); + if (error) + return error; + dentry = nd->dentry; +@@ -1043,6 +1159,10 @@ + /* + * Create - we need to know the parent. + */ ++ if (it) { ++ it->it_create_mode = mode; ++ it->it_op |= IT_CREAT; ++ } + error = path_lookup(pathname, LOOKUP_PARENT, nd); + if (error) + return error; +@@ -1058,7 +1178,7 @@ + + dir = nd->dentry; + down(&dir->d_inode->i_sem); +- dentry = lookup_hash(&nd->last, nd->dentry); ++ dentry = lookup_hash_it(&nd->last, nd->dentry, it); + + do_last: + error = PTR_ERR(dentry); +@@ -1067,11 +1187,12 @@ + goto exit; + } + ++ it->it_create_mode = mode; + /* Negative dentry, just create the file */ + if (!dentry->d_inode) { + if (!IS_POSIXACL(dir->d_inode)) + mode &= ~current->fs->umask; +- error = vfs_create(dir->d_inode, dentry, mode); ++ error = vfs_create_it(dir->d_inode, dentry, mode, it); + up(&dir->d_inode->i_sem); + #ifndef DENTRY_WASTE_RAM + if (error) +@@ -1179,7 +1300,7 @@ + if (!error) { + DQUOT_INIT(inode); + +- error = do_truncate(dentry, 0); ++ error = do_truncate(dentry, 0, 1); + } + put_write_access(inode); + if (error) +@@ -1191,8 +1312,10 @@ + return 0; + + exit_dput: ++ intent_release(it); + dput(dentry); + exit: ++ intent_release(it); + path_release(nd); + return error; + +@@ -1211,7 +1334,10 @@ + * are done. Procfs-like symlinks just set LAST_BIND. + */ + UPDATE_ATIME(dentry->d_inode); ++ nd->intent = it; + error = dentry->d_inode->i_op->follow_link(dentry, nd); ++ if (error) ++ intent_release(it); + dput(dentry); + if (error) + return error; +@@ -1233,13 +1359,20 @@ + } + dir = nd->dentry; + down(&dir->d_inode->i_sem); +- dentry = lookup_hash(&nd->last, nd->dentry); ++ dentry = lookup_hash_it(&nd->last, nd->dentry, it); + putname(nd->last.name); + goto do_last; + } + ++int open_namei(const char *pathname, int flag, int mode, struct nameidata *nd) ++{ ++ return open_namei_it(pathname, flag, mode, nd, NULL); ++} ++ ++ + /* SMP-safe */ +-struct dentry *lookup_create(struct nameidata *nd, int is_dir) ++struct dentry *lookup_create(struct nameidata *nd, int is_dir, ++ struct lookup_intent *it) + { + struct dentry *dentry; + +@@ -1247,7 +1380,7 @@ + dentry = ERR_PTR(-EEXIST); + if (nd->last_type != LAST_NORM) + goto fail; +- dentry = lookup_hash(&nd->last, nd->dentry); ++ dentry = lookup_hash_it(&nd->last, nd->dentry, it); + if (IS_ERR(dentry)) + goto fail; + if (!is_dir && nd->last.name[nd->last.len] && !dentry->d_inode) +@@ -1303,7 +1436,16 @@ + error = path_lookup(tmp, LOOKUP_PARENT, &nd); + if (error) + goto out; +- dentry = lookup_create(&nd, 0); ++ ++ if (nd.dentry->d_inode->i_op->mknod_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ error = op->mknod_raw(&nd, mode, dev); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto out2; ++ } ++ ++ dentry = lookup_create(&nd, 0, NULL); + error = PTR_ERR(dentry); + + if (!IS_POSIXACL(nd.dentry->d_inode)) +@@ -1325,6 +1467,7 @@ + dput(dentry); + } + up(&nd.dentry->d_inode->i_sem); ++out2: + path_release(&nd); + out: + putname(tmp); +@@ -1372,7 +1515,14 @@ + error = path_lookup(tmp, LOOKUP_PARENT, &nd); + if (error) + goto out; +- dentry = lookup_create(&nd, 1); ++ if (nd.dentry->d_inode->i_op->mkdir_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ error = op->mkdir_raw(&nd, mode); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto out2; ++ } ++ dentry = lookup_create(&nd, 1, NULL); + error = PTR_ERR(dentry); + if (!IS_ERR(dentry)) { + if (!IS_POSIXACL(nd.dentry->d_inode)) +@@ -1381,6 +1531,7 @@ + dput(dentry); + } + up(&nd.dentry->d_inode->i_sem); ++out2: + path_release(&nd); + out: + putname(tmp); +@@ -1481,8 +1632,16 @@ + error = -EBUSY; + goto exit1; + } ++ if (nd.dentry->d_inode->i_op->rmdir_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ ++ error = op->rmdir_raw(&nd); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto exit1; ++ } + down(&nd.dentry->d_inode->i_sem); +- dentry = lookup_hash(&nd.last, nd.dentry); ++ dentry = lookup_hash_it(&nd.last, nd.dentry, NULL); + error = PTR_ERR(dentry); + if (!IS_ERR(dentry)) { + error = vfs_rmdir(nd.dentry->d_inode, dentry); +@@ -1540,8 +1699,15 @@ + error = -EISDIR; + if (nd.last_type != LAST_NORM) + goto exit1; ++ if (nd.dentry->d_inode->i_op->unlink_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ error = op->unlink_raw(&nd); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto exit1; ++ } + down(&nd.dentry->d_inode->i_sem); +- dentry = lookup_hash(&nd.last, nd.dentry); ++ dentry = lookup_hash_it(&nd.last, nd.dentry, NULL); + error = PTR_ERR(dentry); + if (!IS_ERR(dentry)) { + /* Why not before? Because we want correct error value */ +@@ -1608,15 +1774,23 @@ + error = path_lookup(to, LOOKUP_PARENT, &nd); + if (error) + goto out; +- dentry = lookup_create(&nd, 0); ++ if (nd.dentry->d_inode->i_op->symlink_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ error = op->symlink_raw(&nd, from); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto out2; ++ } ++ dentry = lookup_create(&nd, 0, NULL); + error = PTR_ERR(dentry); + if (!IS_ERR(dentry)) { + error = vfs_symlink(nd.dentry->d_inode, dentry, from); + dput(dentry); + } + up(&nd.dentry->d_inode->i_sem); ++ out2: + path_release(&nd); +-out: ++ out: + putname(to); + } + putname(from); +@@ -1692,7 +1866,14 @@ + error = -EXDEV; + if (old_nd.mnt != nd.mnt) + goto out_release; +- new_dentry = lookup_create(&nd, 0); ++ if (nd.dentry->d_inode->i_op->link_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ error = op->link_raw(&old_nd, &nd); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto out_release; ++ } ++ new_dentry = lookup_create(&nd, 0, NULL); + error = PTR_ERR(new_dentry); + if (!IS_ERR(new_dentry)) { + error = vfs_link(old_nd.dentry, nd.dentry->d_inode, new_dentry); +@@ -1736,7 +1917,7 @@ + * locking]. + */ + int vfs_rename_dir(struct inode *old_dir, struct dentry *old_dentry, +- struct inode *new_dir, struct dentry *new_dentry) ++ struct inode *new_dir, struct dentry *new_dentry) + { + int error; + struct inode *target; +@@ -1815,7 +1996,7 @@ + } + + int vfs_rename_other(struct inode *old_dir, struct dentry *old_dentry, +- struct inode *new_dir, struct dentry *new_dentry) ++ struct inode *new_dir, struct dentry *new_dentry) + { + int error; + +@@ -1903,9 +2084,18 @@ + if (newnd.last_type != LAST_NORM) + goto exit2; + ++ if (old_dir->d_inode->i_op->rename_raw) { ++ lock_kernel(); ++ error = old_dir->d_inode->i_op->rename_raw(&oldnd, &newnd); ++ unlock_kernel(); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto exit2; ++ } ++ + double_lock(new_dir, old_dir); + +- old_dentry = lookup_hash(&oldnd.last, old_dir); ++ old_dentry = lookup_hash_it(&oldnd.last, old_dir, NULL); + error = PTR_ERR(old_dentry); + if (IS_ERR(old_dentry)) + goto exit3; +@@ -1921,16 +2111,16 @@ + if (newnd.last.name[newnd.last.len]) + goto exit4; + } +- new_dentry = lookup_hash(&newnd.last, new_dir); ++ new_dentry = lookup_hash_it(&newnd.last, new_dir, NULL); + error = PTR_ERR(new_dentry); + if (IS_ERR(new_dentry)) + goto exit4; + ++ + lock_kernel(); + error = vfs_rename(old_dir->d_inode, old_dentry, + new_dir->d_inode, new_dentry); + unlock_kernel(); +- + dput(new_dentry); + exit4: + dput(old_dentry); +@@ -1981,20 +2171,26 @@ + } + + static inline int __attribute__((always_inline)) +-__vfs_follow_link(struct nameidata *nd, const char *link) ++__vfs_follow_link(struct nameidata *nd, const char *link, ++ struct lookup_intent *it) + { + int res = 0; + char *name; + if (IS_ERR(link)) + goto fail; + ++ if (it == NULL) ++ it = nd->intent; ++ else if (it != nd->intent) ++ printk("it != nd->intent: tell phil@clusterfs.com\n"); ++ + if (*link == '/') { + path_release(nd); + if (!walk_init_root(link, nd)) + /* weird __emul_prefix() stuff did it */ + goto out; + } +- res = __link_path_walk(link, nd); ++ res = __link_path_walk_it(link, nd, it); + out: + if (current->link_count || res || nd->last_type!=LAST_NORM) + return res; +@@ -2018,7 +2214,13 @@ + + int vfs_follow_link(struct nameidata *nd, const char *link) + { +- return __vfs_follow_link(nd, link); ++ return __vfs_follow_link(nd, link, NULL); ++} ++ ++int vfs_follow_link_it(struct nameidata *nd, const char *link, ++ struct lookup_intent *it) ++{ ++ return __vfs_follow_link(nd, link, it); + } + + /* get the link contents into pagecache */ +@@ -2060,7 +2262,7 @@ + { + struct page *page = NULL; + char *s = page_getlink(dentry, &page); +- int res = __vfs_follow_link(nd, s); ++ int res = __vfs_follow_link(nd, s, NULL); + if (page) { + kunmap(page); + page_cache_release(page); +Index: linux-2.4.21-171/fs/namespace.c +=================================================================== +--- linux-2.4.21-171.orig/fs/namespace.c 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/fs/namespace.c 2004-04-03 16:09:50.000000000 -0500 +@@ -98,6 +98,7 @@ + { + old_nd->dentry = mnt->mnt_mountpoint; + old_nd->mnt = mnt->mnt_parent; ++ UNPIN(old_nd->dentry, old_nd->mnt, 1); + mnt->mnt_parent = mnt; + mnt->mnt_mountpoint = mnt->mnt_root; + list_del_init(&mnt->mnt_child); +@@ -109,6 +110,7 @@ + { + mnt->mnt_parent = mntget(nd->mnt); + mnt->mnt_mountpoint = dget(nd->dentry); ++ PIN(nd->dentry, nd->mnt, 1); + list_add(&mnt->mnt_hash, mount_hashtable+hash(nd->mnt, nd->dentry)); + list_add(&mnt->mnt_child, &nd->mnt->mnt_mounts); + nd->dentry->d_mounted++; +@@ -488,14 +490,17 @@ + { + struct nameidata old_nd; + struct vfsmount *mnt = NULL; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + int err = mount_is_safe(nd); + if (err) + return err; + if (!old_name || !*old_name) + return -EINVAL; +- err = path_lookup(old_name, LOOKUP_POSITIVE|LOOKUP_FOLLOW, &old_nd); +- if (err) ++ err = path_lookup_it(old_name, LOOKUP_POSITIVE|LOOKUP_FOLLOW, &old_nd, &it); ++ if (err) { ++ intent_release(&it); + return err; ++ } + + down_write(¤t->namespace->sem); + err = -EINVAL; +@@ -518,6 +523,7 @@ + } + + up_write(¤t->namespace->sem); ++ intent_release(&it); + path_release(&old_nd); + return err; + } +@@ -701,6 +707,7 @@ + unsigned long flags, void *data_page) + { + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + int retval = 0; + int mnt_flags = 0; + +@@ -725,10 +732,11 @@ + flags &= ~(MS_NOSUID|MS_NOEXEC|MS_NODEV); + + /* ... and get the mountpoint */ +- retval = path_lookup(dir_name, LOOKUP_FOLLOW|LOOKUP_POSITIVE, &nd); +- if (retval) ++ retval = path_lookup_it(dir_name, LOOKUP_FOLLOW|LOOKUP_POSITIVE, &nd, &it); ++ if (retval) { ++ intent_release(&it); + return retval; +- ++ } + if (flags & MS_REMOUNT) + retval = do_remount(&nd, flags & ~MS_REMOUNT, mnt_flags, + data_page); +@@ -739,6 +747,8 @@ + else + retval = do_add_mount(&nd, type_page, flags, mnt_flags, + dev_name, data_page); ++ ++ intent_release(&it); + path_release(&nd); + return retval; + } +@@ -904,6 +914,8 @@ + { + struct vfsmount *tmp; + struct nameidata new_nd, old_nd, parent_nd, root_parent, user_nd; ++ struct lookup_intent new_it = { .it_op = IT_GETATTR }; ++ struct lookup_intent old_it = { .it_op = IT_GETATTR }; + int error; + + if (!capable(CAP_SYS_ADMIN)) +@@ -911,14 +923,14 @@ + + lock_kernel(); + +- error = __user_walk(new_root, LOOKUP_POSITIVE|LOOKUP_FOLLOW|LOOKUP_DIRECTORY, &new_nd); ++ error = __user_walk_it(new_root, LOOKUP_POSITIVE|LOOKUP_FOLLOW|LOOKUP_DIRECTORY, &new_nd, &new_it); + if (error) + goto out0; + error = -EINVAL; + if (!check_mnt(new_nd.mnt)) + goto out1; + +- error = __user_walk(put_old, LOOKUP_POSITIVE|LOOKUP_FOLLOW|LOOKUP_DIRECTORY, &old_nd); ++ error = __user_walk_it(put_old, LOOKUP_POSITIVE|LOOKUP_FOLLOW|LOOKUP_DIRECTORY, &old_nd, &old_it); + if (error) + goto out1; + +@@ -973,8 +985,10 @@ + up(&old_nd.dentry->d_inode->i_zombie); + up_write(¤t->namespace->sem); + path_release(&user_nd); ++ intent_release(&old_it); + path_release(&old_nd); + out1: ++ intent_release(&new_it); + path_release(&new_nd); + out0: + unlock_kernel(); +Index: linux-2.4.21-171/fs/open.c +=================================================================== +--- linux-2.4.21-171.orig/fs/open.c 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/fs/open.c 2004-04-03 16:09:50.000000000 -0500 +@@ -19,6 +19,8 @@ + #include + + #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) ++extern int path_walk_it(const char *name, struct nameidata *nd, ++ struct lookup_intent *it); + + int vfs_statfs(struct super_block *sb, struct statfs *buf) + { +@@ -95,9 +97,10 @@ + write_unlock(&files->file_lock); + } + +-int do_truncate(struct dentry *dentry, loff_t length) ++int do_truncate(struct dentry *dentry, loff_t length, int called_from_open) + { + struct inode *inode = dentry->d_inode; ++ struct inode_operations *op = dentry->d_inode->i_op; + int error; + struct iattr newattrs; + +@@ -109,7 +112,13 @@ + down(&inode->i_sem); + newattrs.ia_size = length; + newattrs.ia_valid = ATTR_SIZE | ATTR_CTIME; +- error = notify_change(dentry, &newattrs); ++ if (called_from_open) ++ newattrs.ia_valid |= ATTR_FROM_OPEN; ++ if (op->setattr_raw) { ++ newattrs.ia_valid |= ATTR_RAW; ++ error = op->setattr_raw(inode, &newattrs); ++ } else ++ error = notify_change(dentry, &newattrs); + up(&inode->i_sem); + up_write(&inode->i_alloc_sem); + return error; +@@ -120,12 +129,13 @@ + struct nameidata nd; + struct inode * inode; + int error; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + + error = -EINVAL; + if (length < 0) /* sorry, but loff_t says... */ + goto out; + +- error = user_path_walk(path, &nd); ++ error = user_path_walk_it(path, &nd, &it); + if (error) + goto out; + inode = nd.dentry->d_inode; +@@ -165,11 +175,13 @@ + error = locks_verify_truncate(inode, NULL, length); + if (!error) { + DQUOT_INIT(inode); +- error = do_truncate(nd.dentry, length); ++ intent_release(&it); ++ error = do_truncate(nd.dentry, length, 0); + } + put_write_access(inode); + + dput_and_out: ++ intent_release(&it); + path_release(&nd); + out: + return error; +@@ -217,7 +229,7 @@ + + error = locks_verify_truncate(inode, file, length); + if (!error) +- error = do_truncate(dentry, length); ++ error = do_truncate(dentry, length, 0); + out_putf: + fput(file); + out: +@@ -262,11 +274,13 @@ + struct inode * inode; + struct iattr newattrs; + +- error = user_path_walk(filename, &nd); ++ error = user_path_walk_it(filename, &nd, NULL); + if (error) + goto out; + inode = nd.dentry->d_inode; + ++ /* this is safe without a Lustre lock because it only depends ++ on the super block */ + error = -EROFS; + if (IS_RDONLY(inode)) + goto dput_and_out; +@@ -281,11 +295,25 @@ + goto dput_and_out; + + newattrs.ia_valid |= ATTR_ATIME_SET | ATTR_MTIME_SET; +- } else { ++ } ++ ++ if (inode->i_op->setattr_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ ++ newattrs.ia_valid |= ATTR_RAW; ++ error = op->setattr_raw(inode, &newattrs); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto dput_and_out; ++ } ++ ++ error = -EPERM; ++ if (!times) { + if (current->fsuid != inode->i_uid && + (error = permission(inode,MAY_WRITE)) != 0) + goto dput_and_out; + } ++ + error = notify_change(nd.dentry, &newattrs); + dput_and_out: + path_release(&nd); +@@ -306,12 +334,14 @@ + struct inode * inode; + struct iattr newattrs; + +- error = user_path_walk(filename, &nd); ++ error = user_path_walk_it(filename, &nd, NULL); + + if (error) + goto out; + inode = nd.dentry->d_inode; + ++ /* this is safe without a Lustre lock because it only depends ++ on the super block */ + error = -EROFS; + if (IS_RDONLY(inode)) + goto dput_and_out; +@@ -326,7 +356,20 @@ + newattrs.ia_atime = times[0].tv_sec; + newattrs.ia_mtime = times[1].tv_sec; + newattrs.ia_valid |= ATTR_ATIME_SET | ATTR_MTIME_SET; +- } else { ++ } ++ ++ if (inode->i_op->setattr_raw) { ++ struct inode_operations *op = nd.dentry->d_inode->i_op; ++ ++ newattrs.ia_valid |= ATTR_RAW; ++ error = op->setattr_raw(inode, &newattrs); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ goto dput_and_out; ++ } ++ ++ error = -EPERM; ++ if (!utimes) { + if (current->fsuid != inode->i_uid && + (error = permission(inode,MAY_WRITE)) != 0) + goto dput_and_out; +@@ -349,6 +392,7 @@ + int old_fsuid, old_fsgid; + kernel_cap_t old_cap; + int res; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + + if (mode & ~S_IRWXO) /* where's F_OK, X_OK, W_OK, R_OK? */ + return -EINVAL; +@@ -366,13 +410,14 @@ + else + current->cap_effective = current->cap_permitted; + +- res = user_path_walk(filename, &nd); ++ res = user_path_walk_it(filename, &nd, &it); + if (!res) { + res = permission(nd.dentry->d_inode, mode); + /* SuS v2 requires we report a read only fs too */ + if(!res && (mode & S_IWOTH) && IS_RDONLY(nd.dentry->d_inode) + && !special_file(nd.dentry->d_inode->i_mode)) + res = -EROFS; ++ intent_release(&it); + path_release(&nd); + } + +@@ -387,8 +432,9 @@ + { + int error; + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + +- error = __user_walk(filename,LOOKUP_POSITIVE|LOOKUP_FOLLOW|LOOKUP_DIRECTORY,&nd); ++ error = __user_walk_it(filename,LOOKUP_POSITIVE|LOOKUP_FOLLOW|LOOKUP_DIRECTORY,&nd, &it); + if (error) + goto out; + +@@ -399,6 +445,7 @@ + set_fs_pwd(current->fs, nd.mnt, nd.dentry); + + dput_and_out: ++ intent_release(&it); + path_release(&nd); + out: + return error; +@@ -438,9 +485,10 @@ + { + int error; + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + +- error = __user_walk(filename, LOOKUP_POSITIVE | LOOKUP_FOLLOW | +- LOOKUP_DIRECTORY | LOOKUP_NOALT, &nd); ++ error = __user_walk_it(filename, LOOKUP_POSITIVE | LOOKUP_FOLLOW | ++ LOOKUP_DIRECTORY | LOOKUP_NOALT, &nd, &it); + if (error) + goto out; + +@@ -456,39 +504,56 @@ + set_fs_altroot(); + error = 0; + dput_and_out: ++ intent_release(&it); + path_release(&nd); + out: + return error; + } + +-asmlinkage long sys_fchmod(unsigned int fd, mode_t mode) ++int chmod_common(struct dentry *dentry, mode_t mode) + { +- struct inode * inode; +- struct dentry * dentry; +- struct file * file; +- int err = -EBADF; ++ struct inode *inode = dentry->d_inode; + struct iattr newattrs; ++ int err = -EROFS; + +- file = fget(fd); +- if (!file) ++ if (IS_RDONLY(inode)) + goto out; + +- dentry = file->f_dentry; +- inode = dentry->d_inode; ++ if (inode->i_op->setattr_raw) { ++ newattrs.ia_mode = mode; ++ newattrs.ia_valid = ATTR_MODE | ATTR_CTIME; ++ newattrs.ia_valid |= ATTR_RAW; ++ err = inode->i_op->setattr_raw(inode, &newattrs); ++ /* the file system wants to use normal vfs path now */ ++ if (err != -EOPNOTSUPP) ++ goto out; ++ } + +- err = -EROFS; +- if (IS_RDONLY(inode)) +- goto out_putf; + err = -EPERM; + if (IS_IMMUTABLE(inode) || IS_APPEND(inode)) +- goto out_putf; ++ goto out; ++ + if (mode == (mode_t) -1) + mode = inode->i_mode; + newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO); + newattrs.ia_valid = ATTR_MODE | ATTR_CTIME; + err = notify_change(dentry, &newattrs); + +-out_putf: ++out: ++ return err; ++} ++ ++asmlinkage long sys_fchmod(unsigned int fd, mode_t mode) ++{ ++ struct file * file; ++ int err = -EBADF; ++ ++ file = fget(fd); ++ if (!file) ++ goto out; ++ ++ err = chmod_common(file->f_dentry, mode); ++ + fput(file); + out: + return err; +@@ -497,30 +562,14 @@ + asmlinkage long sys_chmod(const char * filename, mode_t mode) + { + struct nameidata nd; +- struct inode * inode; + int error; +- struct iattr newattrs; + + error = user_path_walk(filename, &nd); + if (error) + goto out; +- inode = nd.dentry->d_inode; +- +- error = -EROFS; +- if (IS_RDONLY(inode)) +- goto dput_and_out; + +- error = -EPERM; +- if (IS_IMMUTABLE(inode) || IS_APPEND(inode)) +- goto dput_and_out; ++ error = chmod_common(nd.dentry, mode); + +- if (mode == (mode_t) -1) +- mode = inode->i_mode; +- newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO); +- newattrs.ia_valid = ATTR_MODE | ATTR_CTIME; +- error = notify_change(nd.dentry, &newattrs); +- +-dput_and_out: + path_release(&nd); + out: + return error; +@@ -540,6 +589,20 @@ + error = -EROFS; + if (IS_RDONLY(inode)) + goto out; ++ ++ if (inode->i_op->setattr_raw) { ++ struct inode_operations *op = dentry->d_inode->i_op; ++ ++ newattrs.ia_uid = user; ++ newattrs.ia_gid = group; ++ newattrs.ia_valid = ATTR_UID | ATTR_GID | ATTR_CTIME; ++ newattrs.ia_valid |= ATTR_RAW; ++ error = op->setattr_raw(inode, &newattrs); ++ /* the file system wants to use normal vfs path now */ ++ if (error != -EOPNOTSUPP) ++ return error; ++ } ++ + error = -EPERM; + if (IS_IMMUTABLE(inode) || IS_APPEND(inode)) + goto out; +@@ -644,6 +707,7 @@ + { + int namei_flags, error; + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_OPEN }; + + namei_flags = flags; + if ((namei_flags+1) & O_ACCMODE) +@@ -651,14 +715,15 @@ + if (namei_flags & O_TRUNC) + namei_flags |= 2; + +- error = open_namei(filename, namei_flags, mode, &nd); +- if (!error) +- return dentry_open(nd.dentry, nd.mnt, flags); ++ error = open_namei_it(filename, namei_flags, mode, &nd, &it); ++ if (error) ++ return ERR_PTR(error); + +- return ERR_PTR(error); ++ return dentry_open_it(nd.dentry, nd.mnt, flags, &it); + } + +-struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags) ++struct file *dentry_open_it(struct dentry *dentry, struct vfsmount *mnt, ++ int flags, struct lookup_intent *it) + { + struct file * f; + struct inode *inode; +@@ -686,7 +751,9 @@ + file_move(f, &inode->i_sb->s_files); + + if (f->f_op && f->f_op->open) { ++ f->f_it = it; + error = f->f_op->open(inode,f); ++ f->f_it = NULL; + if (error) + goto cleanup_all; + } +@@ -698,6 +765,7 @@ + !inode->i_mapping->a_ops->direct_IO)) + goto cleanup_all; + ++ intent_release(it); + return f; + + cleanup_all: +@@ -710,11 +778,17 @@ + cleanup_file: + put_filp(f); + cleanup_dentry: ++ intent_release(it); + dput(dentry); + mntput(mnt); + return ERR_PTR(error); + } + ++struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags) ++{ ++ return dentry_open_it(dentry, mnt, flags, NULL); ++} ++ + /* + * Find an empty file descriptor entry, and mark it busy. + */ +Index: linux-2.4.21-171/fs/stat.c +=================================================================== +--- linux-2.4.21-171.orig/fs/stat.c 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/fs/stat.c 2004-04-03 16:09:50.000000000 -0500 +@@ -17,10 +17,14 @@ + * Revalidate the inode. This is required for proper NFS attribute caching. + */ + static __inline__ int +-do_revalidate(struct dentry *dentry) ++do_revalidate(struct dentry *dentry, struct lookup_intent *it) + { + struct inode * inode = dentry->d_inode; +- if (inode->i_op && inode->i_op->revalidate) ++ if (!inode) ++ return -ENOENT; ++ if (inode->i_op && inode->i_op->revalidate_it) ++ return inode->i_op->revalidate_it(dentry, it); ++ else if (inode->i_op && inode->i_op->revalidate) + return inode->i_op->revalidate(dentry); + return 0; + } +@@ -141,13 +145,15 @@ + asmlinkage long sys_stat(char * filename, struct __old_kernel_stat * statbuf) + { + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + int error; + +- error = user_path_walk(filename, &nd); ++ error = user_path_walk_it(filename, &nd, &it); + if (!error) { +- error = do_revalidate(nd.dentry); ++ error = do_revalidate(nd.dentry, &it); + if (!error) + error = cp_old_stat(nd.dentry->d_inode, statbuf); ++ intent_release(&it); + path_release(&nd); + } + return error; +@@ -157,13 +163,15 @@ + asmlinkage long sys_newstat(char * filename, struct stat * statbuf) + { + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + int error; + +- error = user_path_walk(filename, &nd); ++ error = user_path_walk_it(filename, &nd, &it); + if (!error) { +- error = do_revalidate(nd.dentry); ++ error = do_revalidate(nd.dentry, &it); + if (!error) + error = cp_new_stat(nd.dentry->d_inode, statbuf); ++ intent_release(&it); + path_release(&nd); + } + return error; +@@ -178,13 +186,15 @@ + asmlinkage long sys_lstat(char * filename, struct __old_kernel_stat * statbuf) + { + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + int error; + +- error = user_path_walk_link(filename, &nd); ++ error = user_path_walk_link_it(filename, &nd, &it); + if (!error) { +- error = do_revalidate(nd.dentry); ++ error = do_revalidate(nd.dentry, &it); + if (!error) + error = cp_old_stat(nd.dentry->d_inode, statbuf); ++ intent_release(&it); + path_release(&nd); + } + return error; +@@ -195,13 +205,15 @@ + asmlinkage long sys_newlstat(char * filename, struct stat * statbuf) + { + struct nameidata nd; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + int error; + +- error = user_path_walk_link(filename, &nd); ++ error = user_path_walk_link_it(filename, &nd, &it); + if (!error) { +- error = do_revalidate(nd.dentry); ++ error = do_revalidate(nd.dentry, &it); + if (!error) + error = cp_new_stat(nd.dentry->d_inode, statbuf); ++ intent_release(&it); + path_release(&nd); + } + return error; +@@ -222,7 +234,7 @@ + if (f) { + struct dentry * dentry = f->f_dentry; + +- err = do_revalidate(dentry); ++ err = do_revalidate(dentry, NULL); + if (!err) + err = cp_old_stat(dentry->d_inode, statbuf); + fput(f); +@@ -241,7 +253,7 @@ + if (f) { + struct dentry * dentry = f->f_dentry; + +- err = do_revalidate(dentry); ++ err = do_revalidate(dentry, NULL); + if (!err) + err = cp_new_stat(dentry->d_inode, statbuf); + fput(f); +@@ -263,7 +275,7 @@ + + error = -EINVAL; + if (inode->i_op && inode->i_op->readlink && +- !(error = do_revalidate(nd.dentry))) { ++ !(error = do_revalidate(nd.dentry, NULL))) { + UPDATE_ATIME(inode); + error = inode->i_op->readlink(nd.dentry, buf, bufsiz); + } +@@ -339,12 +351,14 @@ + { + struct nameidata nd; + int error; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + +- error = user_path_walk(filename, &nd); ++ error = user_path_walk_it(filename, &nd, &it); + if (!error) { +- error = do_revalidate(nd.dentry); ++ error = do_revalidate(nd.dentry, &it); + if (!error) + error = cp_new_stat64(nd.dentry->d_inode, statbuf); ++ intent_release(&it); + path_release(&nd); + } + return error; +@@ -354,12 +368,14 @@ + { + struct nameidata nd; + int error; ++ struct lookup_intent it = { .it_op = IT_GETATTR }; + +- error = user_path_walk_link(filename, &nd); ++ error = user_path_walk_link_it(filename, &nd, &it); + if (!error) { +- error = do_revalidate(nd.dentry); ++ error = do_revalidate(nd.dentry, &it); + if (!error) + error = cp_new_stat64(nd.dentry->d_inode, statbuf); ++ intent_release(&it); + path_release(&nd); + } + return error; +@@ -374,7 +390,7 @@ + if (f) { + struct dentry * dentry = f->f_dentry; + +- err = do_revalidate(dentry); ++ err = do_revalidate(dentry, NULL); + if (!err) + err = cp_new_stat64(dentry->d_inode, statbuf); + fput(f); +Index: linux-2.4.21-171/include/linux/dcache.h +=================================================================== +--- linux-2.4.21-171.orig/include/linux/dcache.h 2004-03-31 14:58:26.000000000 -0500 ++++ linux-2.4.21-171/include/linux/dcache.h 2004-04-03 16:09:50.000000000 -0500 +@@ -7,6 +7,51 @@ + #include + #include + #include ++#include ++ ++#define IT_OPEN 0x0001 ++#define IT_CREAT 0x0002 ++#define IT_READDIR 0x0004 ++#define IT_GETATTR 0x0008 ++#define IT_LOOKUP 0x0010 ++#define IT_UNLINK 0x0020 ++#define IT_GETXATTR 0x0040 ++#define IT_EXEC 0x0080 ++#define IT_PIN 0x0100 ++ ++#define IT_FL_LOCKED 0x0001 ++#define IT_FL_FOLLOWED 0x0002 /* set by vfs_follow_link */ ++ ++#define INTENT_MAGIC 0x19620323 ++ ++ ++struct lustre_intent_data { ++ int it_disposition; ++ int it_status; ++ __u64 it_lock_handle; ++ void *it_data; ++ int it_lock_mode; ++ int it_int_flags; ++}; ++struct lookup_intent { ++ int it_magic; ++ void (*it_op_release)(struct lookup_intent *); ++ int it_op; ++ int it_flags; ++ int it_create_mode; ++ union { ++ struct lustre_intent_data lustre; ++ } d; ++}; ++ ++static inline void intent_init(struct lookup_intent *it, int op, int flags) ++{ ++ memset(it, 0, sizeof(*it)); ++ it->it_magic = INTENT_MAGIC; ++ it->it_op = op; ++ it->it_flags = flags; ++} ++ + + /* + * linux/include/linux/dcache.h +@@ -94,8 +139,22 @@ + int (*d_delete)(struct dentry *); + void (*d_release)(struct dentry *); + void (*d_iput)(struct dentry *, struct inode *); ++ int (*d_revalidate_it)(struct dentry *, int, struct lookup_intent *); ++ void (*d_pin)(struct dentry *, struct vfsmount * , int); ++ void (*d_unpin)(struct dentry *, struct vfsmount *, int); + }; + ++#define PIN(de,mnt,flag) if (de && de->d_op && de->d_op->d_pin) \ ++ de->d_op->d_pin(de, mnt, flag); ++#define UNPIN(de,mnt,flag) if (de && de->d_op && de->d_op->d_unpin) \ ++ de->d_op->d_unpin(de, mnt, flag); ++ ++ ++/* defined in fs/namei.c */ ++extern void intent_release(struct lookup_intent *it); ++/* defined in fs/dcache.c */ ++extern void __d_rehash(struct dentry * entry, int lock); ++ + /* the dentry parameter passed to d_hash and d_compare is the parent + * directory of the entries to be compared. It is used in case these + * functions need any directory specific information for determining +@@ -127,6 +186,7 @@ + * s_nfsd_free_path semaphore will be down + */ + #define DCACHE_REFERENCED 0x0008 /* Recently used, don't discard. */ ++#define DCACHE_LUSTRE_INVALID 0x0010 /* Lustre invalidated */ + + extern spinlock_t dcache_lock; + +Index: linux-2.4.21-171/include/linux/fs.h +=================================================================== +--- linux-2.4.21-171.orig/include/linux/fs.h 2004-04-03 16:07:40.000000000 -0500 ++++ linux-2.4.21-171/include/linux/fs.h 2004-04-03 16:09:50.000000000 -0500 +@@ -74,6 +74,7 @@ + + #define FMODE_READ 1 + #define FMODE_WRITE 2 ++#define FMODE_EXEC 4 + + #define READ 0 + #define WRITE 1 +@@ -360,6 +361,9 @@ + #define ATTR_MTIME_SET 256 + #define ATTR_FORCE 512 /* Not a change, but a change it */ + #define ATTR_ATTR_FLAG 1024 ++#define ATTR_RAW 0x0800 /* file system, not vfs will massage attrs */ ++#define ATTR_FROM_OPEN 0x1000 /* called from open path, ie O_TRUNC */ ++#define ATTR_CTIME_SET 0x2000 + + /* + * This is the Inode Attributes structure, used for notify_change(). It +@@ -504,6 +508,7 @@ + struct pipe_inode_info *i_pipe; + struct block_device *i_bdev; + struct char_device *i_cdev; ++ void *i_filterdata; + + unsigned long i_dnotify_mask; /* Directory notify events */ + struct dnotify_struct *i_dnotify; /* for directory notifications */ +@@ -666,6 +671,7 @@ + + /* needed for tty driver, and maybe others */ + void *private_data; ++ struct lookup_intent *f_it; + struct list_head f_ep_links; + spinlock_t f_ep_lock; + }; +@@ -795,6 +801,7 @@ + struct qstr last; + unsigned int flags; + int last_type; ++ struct lookup_intent *intent; + }; + + /* +@@ -916,7 +923,8 @@ + extern int __vfs_rmdir(struct inode *, struct dentry *); + extern int vfs_rmdir(struct inode *, struct dentry *); + extern int vfs_unlink(struct inode *, struct dentry *); +-extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); ++int vfs_rename(struct inode *old_dir, struct dentry *old_dentry, ++ struct inode *new_dir, struct dentry *new_dentry); + + /* + * File types +@@ -991,21 +999,32 @@ + + struct inode_operations { + int (*create) (struct inode *,struct dentry *,int); ++ int (*create_it) (struct inode *,struct dentry *,int, struct lookup_intent *); + struct dentry * (*lookup) (struct inode *,struct dentry *); ++ struct dentry * (*lookup_it) (struct inode *,struct dentry *, struct lookup_intent *, int flags); + int (*link) (struct dentry *,struct inode *,struct dentry *); ++ int (*link_raw) (struct nameidata *,struct nameidata *); + int (*unlink) (struct inode *,struct dentry *); ++ int (*unlink_raw) (struct nameidata *); + int (*symlink) (struct inode *,struct dentry *,const char *); ++ int (*symlink_raw) (struct nameidata *,const char *); + int (*mkdir) (struct inode *,struct dentry *,int); ++ int (*mkdir_raw) (struct nameidata *,int); + int (*rmdir) (struct inode *,struct dentry *); ++ int (*rmdir_raw) (struct nameidata *); + int (*mknod) (struct inode *,struct dentry *,int,int); ++ int (*mknod_raw) (struct nameidata *,int,dev_t); + int (*rename) (struct inode *, struct dentry *, + struct inode *, struct dentry *); ++ int (*rename_raw) (struct nameidata *, struct nameidata *); + int (*readlink) (struct dentry *, char *,int); + int (*follow_link) (struct dentry *, struct nameidata *); + void (*truncate) (struct inode *); + int (*permission) (struct inode *, int); + int (*revalidate) (struct dentry *); ++ int (*revalidate_it) (struct dentry *, struct lookup_intent *); + int (*setattr) (struct dentry *, struct iattr *); ++ int (*setattr_raw) (struct inode *, struct iattr *); + int (*getattr) (struct dentry *, struct iattr *); + int (*setxattr) (struct dentry *, const char *, const void *, size_t, int); + ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); +@@ -1204,10 +1223,14 @@ + + asmlinkage long sys_open(const char *, int, int); + asmlinkage long sys_close(unsigned int); /* yes, it's really unsigned */ +-extern int do_truncate(struct dentry *, loff_t start); ++extern int do_truncate(struct dentry *, loff_t start, int called_from_open); + + extern struct file *filp_open(const char *, int, int); + extern struct file * dentry_open(struct dentry *, struct vfsmount *, int); ++extern int open_namei_it(const char *filename, int namei_flags, int mode, ++ struct nameidata *nd, struct lookup_intent *it); ++extern struct file *dentry_open_it(struct dentry *dentry, struct vfsmount *mnt, ++ int flags, struct lookup_intent *it); + extern int filp_close(struct file *, fl_owner_t id); + extern char * getname(const char *); + +@@ -1503,6 +1526,7 @@ + extern loff_t default_llseek(struct file *file, loff_t offset, int origin); + + extern int FASTCALL(__user_walk(const char *, unsigned, struct nameidata *)); ++extern int FASTCALL(__user_walk_it(const char *, unsigned, struct nameidata *, struct lookup_intent *it)); + extern int FASTCALL(path_init(const char *, unsigned, struct nameidata *)); + extern int FASTCALL(path_walk(const char *, struct nameidata *)); + extern int FASTCALL(path_lookup(const char *, unsigned, struct nameidata *)); +@@ -1515,6 +1539,8 @@ + extern struct dentry * lookup_hash(struct qstr *, struct dentry *); + #define user_path_walk(name,nd) __user_walk(name, LOOKUP_FOLLOW|LOOKUP_POSITIVE, nd) + #define user_path_walk_link(name,nd) __user_walk(name, LOOKUP_POSITIVE, nd) ++#define user_path_walk_it(name,nd,it) __user_walk_it(name, LOOKUP_FOLLOW|LOOKUP_POSITIVE, nd, it) ++#define user_path_walk_link_it(name,nd,it) __user_walk_it(name, LOOKUP_POSITIVE, nd, it) + + extern void inode_init_once(struct inode *); + extern void _inode_init_once(struct inode *); +@@ -1666,6 +1692,8 @@ + + extern int vfs_readlink(struct dentry *, char *, int, const char *); + extern int vfs_follow_link(struct nameidata *, const char *); ++extern int vfs_follow_link_it(struct nameidata *, const char *, ++ struct lookup_intent *it); + extern int page_readlink(struct dentry *, char *, int); + extern int page_follow_link(struct dentry *, struct nameidata *); + extern struct inode_operations page_symlink_inode_operations; +Index: linux-2.4.21-171/include/linux/fs_struct.h +=================================================================== +--- linux-2.4.21-171.orig/include/linux/fs_struct.h 2004-02-24 13:42:29.000000000 -0500 ++++ linux-2.4.21-171/include/linux/fs_struct.h 2004-04-03 16:09:50.000000000 -0500 +@@ -34,10 +34,12 @@ + write_lock(&fs->lock); + old_root = fs->root; + old_rootmnt = fs->rootmnt; ++ PIN(dentry, mnt, 1); + fs->rootmnt = mntget(mnt); + fs->root = dget(dentry); + write_unlock(&fs->lock); + if (old_root) { ++ UNPIN(old_root, old_rootmnt, 1); + dput(old_root); + mntput(old_rootmnt); + } +@@ -57,10 +59,12 @@ + write_lock(&fs->lock); + old_pwd = fs->pwd; + old_pwdmnt = fs->pwdmnt; ++ PIN(dentry, mnt, 0); + fs->pwdmnt = mntget(mnt); + fs->pwd = dget(dentry); + write_unlock(&fs->lock); + if (old_pwd) { ++ UNPIN(old_pwd, old_pwdmnt, 0); + dput(old_pwd); + mntput(old_pwdmnt); + } +Index: linux-2.4.21-171/kernel/exit.c +=================================================================== +--- linux-2.4.21-171.orig/kernel/exit.c 2004-02-24 14:03:21.000000000 -0500 ++++ linux-2.4.21-171/kernel/exit.c 2004-04-03 16:09:50.000000000 -0500 +@@ -292,11 +292,14 @@ + { + /* No need to hold fs->lock if we are killing it */ + if (atomic_dec_and_test(&fs->count)) { ++ UNPIN(fs->pwd, fs->pwdmnt, 0); ++ UNPIN(fs->root, fs->rootmnt, 1); + dput(fs->root); + mntput(fs->rootmnt); + dput(fs->pwd); + mntput(fs->pwdmnt); + if (fs->altroot) { ++ UNPIN(fs->altroot, fs->altrootmnt, 1); + dput(fs->altroot); + mntput(fs->altrootmnt); + } +Index: linux-2.4.21-171/kernel/fork.c +=================================================================== +--- linux-2.4.21-171.orig/kernel/fork.c 2004-02-24 14:03:21.000000000 -0500 ++++ linux-2.4.21-171/kernel/fork.c 2004-04-03 16:09:50.000000000 -0500 +@@ -467,10 +467,13 @@ + fs->umask = old->umask; + read_lock(&old->lock); + fs->rootmnt = mntget(old->rootmnt); ++ PIN(old->pwd, old->pwdmnt, 0); ++ PIN(old->root, old->rootmnt, 1); + fs->root = dget(old->root); + fs->pwdmnt = mntget(old->pwdmnt); + fs->pwd = dget(old->pwd); + if (old->altroot) { ++ PIN(old->altroot, old->altrootmnt, 1); + fs->altrootmnt = mntget(old->altrootmnt); + fs->altroot = dget(old->altroot); + } else { +Index: linux-2.4.21-171/kernel/ksyms.c +=================================================================== +--- linux-2.4.21-171.orig/kernel/ksyms.c 2004-04-03 16:07:40.000000000 -0500 ++++ linux-2.4.21-171/kernel/ksyms.c 2004-04-03 16:09:50.000000000 -0500 +@@ -334,6 +334,7 @@ + EXPORT_SYMBOL(set_page_dirty); + EXPORT_SYMBOL(vfs_readlink); + EXPORT_SYMBOL(vfs_follow_link); ++EXPORT_SYMBOL(vfs_follow_link_it); + EXPORT_SYMBOL(page_readlink); + EXPORT_SYMBOL(page_follow_link); + EXPORT_SYMBOL(page_symlink_inode_operations); diff --git a/lustre/kernel_patches/series/suse-2.4.21-171 b/lustre/kernel_patches/series/suse-2.4.21-171 new file mode 100644 index 0000000..fc59acb --- /dev/null +++ b/lustre/kernel_patches/series/suse-2.4.21-171 @@ -0,0 +1,30 @@ +configurable-x86-stack-2.4.21-suse-171.patch +dev_read_only_2.4.20-rh.patch +exports_2.4.20-rh-hp.patch +lustre_version.patch +vfs_intent-2.4.21-suse-171.patch +invalidate_show.patch +export-truncate.patch +iod-stock-24-exports_hp.patch +ext3-htree-2.4.21-chaos.patch +linux-2.4.21-xattr-0.8.54-suse2.patch +ext3-orphan_lock-2.4.22-rh.patch +ext3-noread-2.4.21-suse2.patch +ext3-delete_thread-2.4.21-chaos.patch +extN-wantedi-2.4.21-suse2.patch +ext3-san-2.4.20.patch +ext3-map_inode_page-2.4.21-suse2.patch +ext3-error-export.patch +iopen-2.4.19-suse.patch +tcp-zero-copy-2.4.21-suse-171.patch +jbd-dont-account-blocks-twice.patch +jbd-commit-tricks.patch +ext3-no-write-super-chaos.patch +add_page_private.patch +nfs_export_kernel-2.4.21-suse2.patch +ext3-raw-lookup.patch +ext3-ea-in-inode-2.4.21-suse2.patch +listman-2.4.20.patch +gfp_memalloc-2.4.24.patch +kernel_text_address-2.4.20-vanilla.patch +ext3-truncate-buffer-head.patch