kernel October 2012

kernel@lists.fedoraproject.org

17 participants
17 discussions

fedora 14 kernel performance with ip forwarding workload
by Jesse Brandeburg 05 Jun '15

05 Jun '15

The other day I was running the stock fedora kernel on my ip forwarding setup, to see what the performance was, and the performance wasn't very good. system is S5520HC dual socket 2.93GHz Xeon 5570 (Nehalem) with 3 quad port 82580 adapters (12 ports). Traffic is bidirectional 64 byte packets being forwarded and received on each port, basically port to port routing. I am only using 12 flows currently. The driver is igb, and I am using an affinity script that lines up each pair of ports that are forwarding traffic into optimal configurations for cache locality. I am also disabling remote_node_defrag_ratio to stop cross node traffic. With the fedora default kernel from F14 it appears that CONFIG_NETFILTER=y means that I cannot unload all of netfilter even if I stop iptables service. perf showed netfilter being prominent, and removing it gives me much higher throughput. Is there a reason CONFIG_NETFILTER=y ? Isn't it a good thing to be able to disable netfilter if you want to? Jesse

5 9

Re: [kernel] Disable debugging options.
by Thorsten Leemhuis 21 Nov '12

21 Nov '12

On 18.06.2012 15:09, Josh Boyer wrote: > commit 3b37beedf48825354c42e25f7001677320958d38 > Author: Josh Boyer <jwboyer(a)redhat.com> > Date: Mon Jun 18 09:09:58 2012 -0400 > > Disable debugging options. > > config-generic | 8 ++-- > config-nodebug | 112 ++++++++++++++++++++++++++-------------------------- > config-x86-generic | 2 +- > kernel.spec | 9 +++- > 4 files changed, 67 insertions(+), 64 deletions(-) > --- > diff --git a/config-generic b/config-generic > index 6b1c651..08c28ba 100644 > --- a/config-generic > +++ b/config-generic > @@ -1446,13 +1446,13 @@ CONFIG_B43_SDIO=y > CONFIG_B43_BCMA=y > # CONFIG_B43_BCMA_EXTRA is not set > CONFIG_B43_BCMA_PIO=y > -CONFIG_B43_DEBUG=y > +# CONFIG_B43_DEBUG is not set > [...] Just out of curiosity: Why is enabling and disabling the debug options done via "make {no,}debug" in the git checkout and not by an conditional within the spec file? That afaics makes it harder to switch the debug options off if you only have the SRPM at hand. Or am I missing something obvious here? CU knurd

3 6

Dropping sparc
by Josh Boyer 31 Oct '12

31 Oct '12

Looking in the sparc koji instance, the last attempt at building a sparc kernel was 2.6.41.10-2.fc15. That was eight months ago. The last attempted official build that I can find of anything else is a hal build from fc13, which failed. Is the Fedora sparc project actually active at all these days or is it something we can finally put to rest? josh

1 0

Building external modules without signing keys?
by Bruno Wolff III 31 Oct '12

31 Oct '12

I build dahdi-linux from source for use on the machine I also run rawhide on and currently I can getting an error message "Can't read private key" during the build. Is there some info on where I should look to work around this? I boot using BIOS, so I don't need to have modules signed.

2 6

[PATCH PULL F17 F18] Uprobes backport from upstream
by Anton Arapov 30 Oct '12

30 Oct '12

Hello, Test builds: f18: http://koji.fedoraproject.org/koji/taskinfo?taskID=4635065 f17: http://koji.fedoraproject.org/koji/taskinfo?taskID=4635062 The split-out series is available in the git repository at: http://fedorapeople.org/cgit/aarapov/public_git/kernel-uprobes.git Just <at> jistone's patch is not upstream that exports functions. Yes, I know it must go to upstream. :-) I will try to do what I can before the next uprobes update. --------------------------------------------------------------- Josh Stone (1): uprobes: add exports necessary for uprobes use by modules Oleg Nesterov (35): uprobes: Kill uprobes_state->count uprobes: Kill dup_mmap()->uprobe_mmap(), simplify uprobe_mmap/munmap uprobes: Change uprobe_mmap() to ignore the errors but check fatal_signal_pending() uprobes: Do not use -EEXIST in install_breakpoint() paths uprobes: Introduce MMF_HAS_UPROBES uprobes: Fold uprobe_reset_state() into uprobe_dup_mmap() uprobes: Remove "verify" argument from set_orig_insn() uprobes: uprobes_treelock should not disable irqs uprobes: Introduce MMF_RECALC_UPROBES uprobes: Teach find_active_uprobe() to clear MMF_HAS_UPROBES ptrace/x86: Introduce set_task_blockstep() helper ptrace/x86: Partly fix set_task_blockstep()->update_debugctlmsr() logic uprobes/x86: Do not (ab)use TIF_SINGLESTEP/user_*_single_step() for single-stepping uprobes/x86: Xol should send SIGTRAP if X86_EFLAGS_TF was set uprobes/x86: Fix arch_uprobe_disable_step() && UTASK_SSTEP_TRAPPED interaction uprobes: Make arch_uprobe_task->saved_trap_nr "unsigned int" uprobes: Do not leak UTASK_BP_HIT if find_active_uprobe() fails uprobes: Do not setup ->active_uprobe/state prematurely uprobes: Fix UPROBE_SKIP_SSTEP checks in handle_swbp() uprobes: Kill UTASK_BP_HIT state uprobes: Move clear_thread_flag(TIF_UPROBE) to uprobe_notify_resume() uprobes: Change write_opcode() to use FOLL_FORCE uprobes: Change valid_vma() to demand VM_MAYEXEC rather than VM_EXEC uprobes: Restrict valid_vma(false) to skip VM_SHARED vmas uprobes: Kill set_swbp()->is_swbp_at_addr() uprobes: Introduce copy_opcode(), kill read_opcode() uprobes: Kill set_orig_insn()->is_swbp_at_addr() uprobes: Simplify is_swbp_at_addr(), remove stale comments uprobes/x86: Only rep+nop can be emulated correctly uprobes: Don't return success if alloc_uprobe() fails uprobes: Do not delete uprobe if uprobe_unregister() fails uprobes: Fix handle_swbp() vs unregister() + register() race uprobes: Introduce prepare_uprobe() uprobes: Fix prepare_uprobe() race with itself uprobes: Fix the racy uprobe->flags manipulation Sebastian Andrzej Siewior (4): uprobes: Remove check for uprobe variable in handle_swbp() uprobes: Don't put NULL pointer in uprobe_register() uprobes: Introduce arch_uprobe_enable/disable_step() uprobes/x86: Implement x86 specific arch_uprobe_*_step Srikar Dronamraju (1): uprobes: Remove redundant lock_page/unlock_page arch/x86/include/asm/processor.h | 2 + arch/x86/include/asm/uprobes.h | 3 +- arch/x86/kernel/signal.c | 4 +- arch/x86/kernel/step.c | 53 ++-- arch/x86/kernel/uprobes.c | 64 ++++- include/linux/sched.h | 3 + include/linux/uprobes.h | 26 +- kernel/events/uprobes.c | 564 ++++++++++++++++++--------------------- kernel/fork.c | 6 +- kernel/ptrace.c | 6 + 10 files changed, 375 insertions(+), 356 deletions(-) diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index d048cad..433d2e5 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -759,6 +759,8 @@ static inline void update_debugctlmsr(unsigned long debugctlmsr) wrmsrl(MSR_IA32_DEBUGCTLMSR, debugctlmsr); } +extern void set_task_blockstep(struct task_struct *task, bool on); + /* * from system description table in BIOS. Mostly for MCA use, but * others may find it useful: diff --git a/arch/x86/include/asm/uprobes.h b/arch/x86/include/asm/uprobes.h index f3971bb..8ff8be7 100644 --- a/arch/x86/include/asm/uprobes.h +++ b/arch/x86/include/asm/uprobes.h @@ -42,10 +42,11 @@ struct arch_uprobe { }; struct arch_uprobe_task { - unsigned long saved_trap_nr; #ifdef CONFIG_X86_64 unsigned long saved_scratch_register; #endif + unsigned int saved_trap_nr; + unsigned int saved_tf; }; extern int arch_uprobe_analyze_insn(struct arch_uprobe *aup, struct mm_struct *mm, unsigned long addr); diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index b280908..0041e5a 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -785,10 +785,8 @@ do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags) mce_notify_process(); #endif /* CONFIG_X86_64 && CONFIG_X86_MCE */ - if (thread_info_flags & _TIF_UPROBE) { - clear_thread_flag(TIF_UPROBE); + if (thread_info_flags & _TIF_UPROBE) uprobe_notify_resume(regs); - } /* deal with pending signal delivery */ if (thread_info_flags & _TIF_SIGPENDING) diff --git a/arch/x86/kernel/step.c b/arch/x86/kernel/step.c index c346d11..cd3b243 100644 --- a/arch/x86/kernel/step.c +++ b/arch/x86/kernel/step.c @@ -157,6 +157,33 @@ static int enable_single_step(struct task_struct *child) return 1; } +void set_task_blockstep(struct task_struct *task, bool on) +{ + unsigned long debugctl; + + /* + * Ensure irq/preemption can't change debugctl in between. + * Note also that both TIF_BLOCKSTEP and debugctl should + * be changed atomically wrt preemption. + * FIXME: this means that set/clear TIF_BLOCKSTEP is simply + * wrong if task != current, SIGKILL can wakeup the stopped + * tracee and set/clear can play with the running task, this + * can confuse the next __switch_to_xtra(). + */ + local_irq_disable(); + debugctl = get_debugctlmsr(); + if (on) { + debugctl |= DEBUGCTLMSR_BTF; + set_tsk_thread_flag(task, TIF_BLOCKSTEP); + } else { + debugctl &= ~DEBUGCTLMSR_BTF; + clear_tsk_thread_flag(task, TIF_BLOCKSTEP); + } + if (task == current) + update_debugctlmsr(debugctl); + local_irq_enable(); +} + /* * Enable single or block step. */ @@ -169,19 +196,10 @@ static void enable_step(struct task_struct *child, bool block) * So no one should try to use debugger block stepping in a program * that uses user-mode single stepping itself. */ - if (enable_single_step(child) && block) { - unsigned long debugctl = get_debugctlmsr(); - - debugctl |= DEBUGCTLMSR_BTF; - update_debugctlmsr(debugctl); - set_tsk_thread_flag(child, TIF_BLOCKSTEP); - } else if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) { - unsigned long debugctl = get_debugctlmsr(); - - debugctl &= ~DEBUGCTLMSR_BTF; - update_debugctlmsr(debugctl); - clear_tsk_thread_flag(child, TIF_BLOCKSTEP); - } + if (enable_single_step(child) && block) + set_task_blockstep(child, true); + else if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) + set_task_blockstep(child, false); } void user_enable_single_step(struct task_struct *child) @@ -199,13 +217,8 @@ void user_disable_single_step(struct task_struct *child) /* * Make sure block stepping (BTF) is disabled. */ - if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) { - unsigned long debugctl = get_debugctlmsr(); - - debugctl &= ~DEBUGCTLMSR_BTF; - update_debugctlmsr(debugctl); - clear_tsk_thread_flag(child, TIF_BLOCKSTEP); - } + if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) + set_task_blockstep(child, false); /* Always clear TIF_SINGLESTEP... */ clear_tsk_thread_flag(child, TIF_SINGLESTEP); diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c index 36fd420..aafa555 100644 --- a/arch/x86/kernel/uprobes.c +++ b/arch/x86/kernel/uprobes.c @@ -41,6 +41,9 @@ /* Adjust the return address of a call insn */ #define UPROBE_FIX_CALL 0x2 +/* Instruction will modify TF, don't change it */ +#define UPROBE_FIX_SETF 0x4 + #define UPROBE_FIX_RIP_AX 0x8000 #define UPROBE_FIX_RIP_CX 0x4000 @@ -239,6 +242,10 @@ static void prepare_fixups(struct arch_uprobe *auprobe, struct insn *insn) insn_get_opcode(insn); /* should be a nop */ switch (OPCODE1(insn)) { + case 0x9d: + /* popf */ + auprobe->fixups |= UPROBE_FIX_SETF; + break; case 0xc3: /* ret/lret */ case 0xcb: case 0xc2: @@ -644,32 +651,63 @@ void arch_uprobe_abort_xol(struct arch_uprobe *auprobe, struct pt_regs *regs) /* * Skip these instructions as per the currently known x86 ISA. - * 0x66* { 0x90 | 0x0f 0x1f | 0x0f 0x19 | 0x87 0xc0 } + * rep=0x66*; nop=0x90 */ -bool arch_uprobe_skip_sstep(struct arch_uprobe *auprobe, struct pt_regs *regs) +static bool __skip_sstep(struct arch_uprobe *auprobe, struct pt_regs *regs) { int i; for (i = 0; i < MAX_UINSN_BYTES; i++) { - if ((auprobe->insn[i] == 0x66)) + if (auprobe->insn[i] == 0x66) continue; if (auprobe->insn[i] == 0x90) return true; - if (i == (MAX_UINSN_BYTES - 1)) - break; + break; + } + return false; +} - if ((auprobe->insn[i] == 0x0f) && (auprobe->insn[i+1] == 0x1f)) - return true; +bool arch_uprobe_skip_sstep(struct arch_uprobe *auprobe, struct pt_regs *regs) +{ + bool ret = __skip_sstep(auprobe, regs); + if (ret && (regs->flags & X86_EFLAGS_TF)) + send_sig(SIGTRAP, current, 0); + return ret; +} - if ((auprobe->insn[i] == 0x0f) && (auprobe->insn[i+1] == 0x19)) - return true; +void arch_uprobe_enable_step(struct arch_uprobe *auprobe) +{ + struct task_struct *task = current; + struct arch_uprobe_task *autask = &task->utask->autask; + struct pt_regs *regs = task_pt_regs(task); - if ((auprobe->insn[i] == 0x87) && (auprobe->insn[i+1] == 0xc0)) - return true; + autask->saved_tf = !!(regs->flags & X86_EFLAGS_TF); - break; + regs->flags |= X86_EFLAGS_TF; + if (test_tsk_thread_flag(task, TIF_BLOCKSTEP)) + set_task_blockstep(task, false); +} + +void arch_uprobe_disable_step(struct arch_uprobe *auprobe) +{ + struct task_struct *task = current; + struct arch_uprobe_task *autask = &task->utask->autask; + bool trapped = (task->utask->state == UTASK_SSTEP_TRAPPED); + struct pt_regs *regs = task_pt_regs(task); + /* + * The state of TIF_BLOCKSTEP was not saved so we can get an extra + * SIGTRAP if we do not clear TF. We need to examine the opcode to + * make it right. + */ + if (unlikely(trapped)) { + if (!autask->saved_tf) + regs->flags &= ~X86_EFLAGS_TF; + } else { + if (autask->saved_tf) + send_sig(SIGTRAP, task, 0); + else if (!(auprobe->fixups & UPROBE_FIX_SETF)) + regs->flags &= ~X86_EFLAGS_TF; } - return false; } diff --git a/include/linux/sched.h b/include/linux/sched.h index 23bddac..ba300be 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -446,6 +446,9 @@ extern int get_dumpable(struct mm_struct *mm); #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */ #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */ +#define MMF_HAS_UPROBES 19 /* has uprobes */ +#define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */ + #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK) struct sighand_struct { diff --git a/include/linux/uprobes.h b/include/linux/uprobes.h index efe4b33..2459457 100644 --- a/include/linux/uprobes.h +++ b/include/linux/uprobes.h @@ -35,16 +35,6 @@ struct inode; # include <asm/uprobes.h> #endif -/* flags that denote/change uprobes behaviour */ - -/* Have a copy of original instruction */ -#define UPROBE_COPY_INSN 0x1 - -/* Dont run handlers when first register/ last unregister in progress*/ -#define UPROBE_RUN_HANDLER 0x2 -/* Can skip singlestep */ -#define UPROBE_SKIP_SSTEP 0x4 - struct uprobe_consumer { int (*handler)(struct uprobe_consumer *self, struct pt_regs *regs); /* @@ -59,7 +49,6 @@ struct uprobe_consumer { #ifdef CONFIG_UPROBES enum uprobe_task_state { UTASK_RUNNING, - UTASK_BP_HIT, UTASK_SSTEP, UTASK_SSTEP_ACK, UTASK_SSTEP_TRAPPED, @@ -99,25 +88,27 @@ struct xol_area { struct uprobes_state { struct xol_area *xol_area; - atomic_t count; }; + extern int __weak set_swbp(struct arch_uprobe *aup, struct mm_struct *mm, unsigned long vaddr); -extern int __weak set_orig_insn(struct arch_uprobe *aup, struct mm_struct *mm, unsigned long vaddr, bool verify); +extern int __weak set_orig_insn(struct arch_uprobe *aup, struct mm_struct *mm, unsigned long vaddr); extern bool __weak is_swbp_insn(uprobe_opcode_t *insn); extern int uprobe_register(struct inode *inode, loff_t offset, struct uprobe_consumer *uc); extern void uprobe_unregister(struct inode *inode, loff_t offset, struct uprobe_consumer *uc); extern int uprobe_mmap(struct vm_area_struct *vma); extern void uprobe_munmap(struct vm_area_struct *vma, unsigned long start, unsigned long end); +extern void uprobe_dup_mmap(struct mm_struct *oldmm, struct mm_struct *newmm); extern void uprobe_free_utask(struct task_struct *t); extern void uprobe_copy_process(struct task_struct *t); extern unsigned long __weak uprobe_get_swbp_addr(struct pt_regs *regs); +extern void __weak arch_uprobe_enable_step(struct arch_uprobe *arch); +extern void __weak arch_uprobe_disable_step(struct arch_uprobe *arch); extern int uprobe_post_sstep_notifier(struct pt_regs *regs); extern int uprobe_pre_sstep_notifier(struct pt_regs *regs); extern void uprobe_notify_resume(struct pt_regs *regs); extern bool uprobe_deny_signal(void); extern bool __weak arch_uprobe_skip_sstep(struct arch_uprobe *aup, struct pt_regs *regs); extern void uprobe_clear_state(struct mm_struct *mm); -extern void uprobe_reset_state(struct mm_struct *mm); #else /* !CONFIG_UPROBES */ struct uprobes_state { }; @@ -138,6 +129,10 @@ static inline void uprobe_munmap(struct vm_area_struct *vma, unsigned long start, unsigned long end) { } +static inline void +uprobe_dup_mmap(struct mm_struct *oldmm, struct mm_struct *newmm) +{ +} static inline void uprobe_notify_resume(struct pt_regs *regs) { } @@ -158,8 +153,5 @@ static inline void uprobe_copy_process(struct task_struct *t) static inline void uprobe_clear_state(struct mm_struct *mm) { } -static inline void uprobe_reset_state(struct mm_struct *mm) -{ -} #endif /* !CONFIG_UPROBES */ #endif /* _LINUX_UPROBES_H */ diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c index c08a22d..e933f65 100644 --- a/kernel/events/uprobes.c +++ b/kernel/events/uprobes.c @@ -27,6 +27,7 @@ #include <linux/pagemap.h> /* read_mapping_page */ #include <linux/slab.h> #include <linux/sched.h> +#include <linux/export.h> #include <linux/rmap.h> /* anon_vma_prepare */ #include <linux/mmu_notifier.h> /* set_pte_at_notify */ #include <linux/swap.h> /* try_to_free_swap */ @@ -78,15 +79,23 @@ static struct mutex uprobes_mmap_mutex[UPROBES_HASH_SZ]; */ static atomic_t uprobe_events = ATOMIC_INIT(0); +/* Have a copy of original instruction */ +#define UPROBE_COPY_INSN 0 +/* Dont run handlers when first register/ last unregister in progress*/ +#define UPROBE_RUN_HANDLER 1 +/* Can skip singlestep */ +#define UPROBE_SKIP_SSTEP 2 + struct uprobe { struct rb_node rb_node; /* node in the rb tree */ atomic_t ref; struct rw_semaphore consumer_rwsem; + struct mutex copy_mutex; /* TODO: kill me and UPROBE_COPY_INSN */ struct list_head pending_list; struct uprobe_consumer *consumers; struct inode *inode; /* Also hold a ref to inode */ loff_t offset; - int flags; + unsigned long flags; struct arch_uprobe arch; }; @@ -100,17 +109,12 @@ struct uprobe { */ static bool valid_vma(struct vm_area_struct *vma, bool is_register) { - if (!vma->vm_file) - return false; + vm_flags_t flags = VM_HUGETLB | VM_MAYEXEC | VM_SHARED; - if (!is_register) - return true; + if (is_register) + flags |= VM_WRITE; - if ((vma->vm_flags & (VM_HUGETLB|VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)) - == (VM_READ|VM_EXEC)) - return true; - - return false; + return vma->vm_file && (vma->vm_flags & flags) == VM_MAYEXEC; } static unsigned long offset_to_vaddr(struct vm_area_struct *vma, loff_t offset) @@ -188,19 +192,44 @@ bool __weak is_swbp_insn(uprobe_opcode_t *insn) return *insn == UPROBE_SWBP_INSN; } +static void copy_opcode(struct page *page, unsigned long vaddr, uprobe_opcode_t *opcode) +{ + void *kaddr = kmap_atomic(page); + memcpy(opcode, kaddr + (vaddr & ~PAGE_MASK), UPROBE_SWBP_INSN_SIZE); + kunmap_atomic(kaddr); +} + +static int verify_opcode(struct page *page, unsigned long vaddr, uprobe_opcode_t *new_opcode) +{ + uprobe_opcode_t old_opcode; + bool is_swbp; + + copy_opcode(page, vaddr, &old_opcode); + is_swbp = is_swbp_insn(&old_opcode); + + if (is_swbp_insn(new_opcode)) { + if (is_swbp) /* register: already installed? */ + return 0; + } else { + if (!is_swbp) /* unregister: was it changed by us? */ + return 0; + } + + return 1; +} + /* * NOTE: * Expect the breakpoint instruction to be the smallest size instruction for * the architecture. If an arch has variable length instruction and the * breakpoint instruction is not of the smallest length instruction - * supported by that architecture then we need to modify read_opcode / + * supported by that architecture then we need to modify is_swbp_at_addr and * write_opcode accordingly. This would never be a problem for archs that * have fixed length instructions. */ /* * write_opcode - write the opcode at a given virtual address. - * @auprobe: arch breakpointing information. * @mm: the probed process address space. * @vaddr: the virtual address to store the opcode. * @opcode: opcode to be written at @vaddr. @@ -211,8 +240,8 @@ bool __weak is_swbp_insn(uprobe_opcode_t *insn) * For mm @mm, write the opcode at @vaddr. * Return 0 (success) or a negative errno. */ -static int write_opcode(struct arch_uprobe *auprobe, struct mm_struct *mm, - unsigned long vaddr, uprobe_opcode_t opcode) +static int write_opcode(struct mm_struct *mm, unsigned long vaddr, + uprobe_opcode_t opcode) { struct page *old_page, *new_page; void *vaddr_old, *vaddr_new; @@ -221,10 +250,14 @@ static int write_opcode(struct arch_uprobe *auprobe, struct mm_struct *mm, retry: /* Read the page with vaddr into memory */ - ret = get_user_pages(NULL, mm, vaddr, 1, 0, 0, &old_page, &vma); + ret = get_user_pages(NULL, mm, vaddr, 1, 0, 1, &old_page, &vma); if (ret <= 0) return ret; + ret = verify_opcode(old_page, vaddr, &opcode); + if (ret <= 0) + goto put_old; + ret = -ENOMEM; new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, vaddr); if (!new_page) @@ -259,65 +292,6 @@ static int write_opcode(struct arch_uprobe *auprobe, struct mm_struct *mm, } /** - * read_opcode - read the opcode at a given virtual address. - * @mm: the probed process address space. - * @vaddr: the virtual address to read the opcode. - * @opcode: location to store the read opcode. - * - * Called with mm->mmap_sem held (for read and with a reference to - * mm. - * - * For mm @mm, read the opcode at @vaddr and store it in @opcode. - * Return 0 (success) or a negative errno. - */ -static int read_opcode(struct mm_struct *mm, unsigned long vaddr, uprobe_opcode_t *opcode) -{ - struct page *page; - void *vaddr_new; - int ret; - - ret = get_user_pages(NULL, mm, vaddr, 1, 0, 1, &page, NULL); - if (ret <= 0) - return ret; - - lock_page(page); - vaddr_new = kmap_atomic(page); - vaddr &= ~PAGE_MASK; - memcpy(opcode, vaddr_new + vaddr, UPROBE_SWBP_INSN_SIZE); - kunmap_atomic(vaddr_new); - unlock_page(page); - - put_page(page); - - return 0; -} - -static int is_swbp_at_addr(struct mm_struct *mm, unsigned long vaddr) -{ - uprobe_opcode_t opcode; - int result; - - if (current->mm == mm) { - pagefault_disable(); - result = __copy_from_user_inatomic(&opcode, (void __user*)vaddr, - sizeof(opcode)); - pagefault_enable(); - - if (likely(result == 0)) - goto out; - } - - result = read_opcode(mm, vaddr, &opcode); - if (result) - return result; -out: - if (is_swbp_insn(&opcode)) - return 1; - - return 0; -} - -/** * set_swbp - store breakpoint at a given address. * @auprobe: arch specific probepoint information. * @mm: the probed process address space. @@ -328,18 +302,7 @@ static int is_swbp_at_addr(struct mm_struct *mm, unsigned long vaddr) */ int __weak set_swbp(struct arch_uprobe *auprobe, struct mm_struct *mm, unsigned long vaddr) { - int result; - /* - * See the comment near uprobes_hash(). - */ - result = is_swbp_at_addr(mm, vaddr); - if (result == 1) - return -EEXIST; - - if (result) - return result; - - return write_opcode(auprobe, mm, vaddr, UPROBE_SWBP_INSN); + return write_opcode(mm, vaddr, UPROBE_SWBP_INSN); } /** @@ -347,25 +310,14 @@ int __weak set_swbp(struct arch_uprobe *auprobe, struct mm_struct *mm, unsigned * @mm: the probed process address space. * @auprobe: arch specific probepoint information. * @vaddr: the virtual address to insert the opcode. - * @verify: if true, verify existance of breakpoint instruction. * * For mm @mm, restore the original opcode (opcode) at @vaddr. * Return 0 (success) or a negative errno. */ int __weak -set_orig_insn(struct arch_uprobe *auprobe, struct mm_struct *mm, unsigned long vaddr, bool verify) +set_orig_insn(struct arch_uprobe *auprobe, struct mm_struct *mm, unsigned long vaddr) { - if (verify) { - int result; - - result = is_swbp_at_addr(mm, vaddr); - if (!result) - return -EINVAL; - - if (result != 1) - return result; - } - return write_opcode(auprobe, mm, vaddr, *(uprobe_opcode_t *)auprobe->insn); + return write_opcode(mm, vaddr, *(uprobe_opcode_t *)auprobe->insn); } static int match_uprobe(struct uprobe *l, struct uprobe *r) @@ -415,11 +367,10 @@ static struct uprobe *__find_uprobe(struct inode *inode, loff_t offset) static struct uprobe *find_uprobe(struct inode *inode, loff_t offset) { struct uprobe *uprobe; - unsigned long flags; - spin_lock_irqsave(&uprobes_treelock, flags); + spin_lock(&uprobes_treelock); uprobe = __find_uprobe(inode, offset); - spin_unlock_irqrestore(&uprobes_treelock, flags); + spin_unlock(&uprobes_treelock); return uprobe; } @@ -466,15 +417,14 @@ static struct uprobe *__insert_uprobe(struct uprobe *uprobe) */ static struct uprobe *insert_uprobe(struct uprobe *uprobe) { - unsigned long flags; struct uprobe *u; - spin_lock_irqsave(&uprobes_treelock, flags); + spin_lock(&uprobes_treelock); u = __insert_uprobe(uprobe); - spin_unlock_irqrestore(&uprobes_treelock, flags); + spin_unlock(&uprobes_treelock); /* For now assume that the instruction need not be single-stepped */ - uprobe->flags |= UPROBE_SKIP_SSTEP; + __set_bit(UPROBE_SKIP_SSTEP, &uprobe->flags); return u; } @@ -496,6 +446,7 @@ static struct uprobe *alloc_uprobe(struct inode *inode, loff_t offset) uprobe->inode = igrab(inode); uprobe->offset = offset; init_rwsem(&uprobe->consumer_rwsem); + mutex_init(&uprobe->copy_mutex); /* add to uprobes_tree, sorted on inode:offset */ cur_uprobe = insert_uprobe(uprobe); @@ -516,7 +467,7 @@ static void handler_chain(struct uprobe *uprobe, struct pt_regs *regs) { struct uprobe_consumer *uc; - if (!(uprobe->flags & UPROBE_RUN_HANDLER)) + if (!test_bit(UPROBE_RUN_HANDLER, &uprobe->flags)) return; down_read(&uprobe->consumer_rwsem); @@ -622,33 +573,48 @@ static int copy_insn(struct uprobe *uprobe, struct file *filp) return __copy_insn(mapping, filp, uprobe->arch.insn, bytes, uprobe->offset); } -/* - * How mm->uprobes_state.count gets updated - * uprobe_mmap() increments the count if - * - it successfully adds a breakpoint. - * - it cannot add a breakpoint, but sees that there is a underlying - * breakpoint (via a is_swbp_at_addr()). - * - * uprobe_munmap() decrements the count if - * - it sees a underlying breakpoint, (via is_swbp_at_addr) - * (Subsequent uprobe_unregister wouldnt find the breakpoint - * unless a uprobe_mmap kicks in, since the old vma would be - * dropped just after uprobe_munmap.) - * - * uprobe_register increments the count if: - * - it successfully adds a breakpoint. - * - * uprobe_unregister decrements the count if: - * - it sees a underlying breakpoint and removes successfully. - * (via is_swbp_at_addr) - * (Subsequent uprobe_munmap wouldnt find the breakpoint - * since there is no underlying breakpoint after the - * breakpoint removal.) - */ +static int prepare_uprobe(struct uprobe *uprobe, struct file *file, + struct mm_struct *mm, unsigned long vaddr) +{ + int ret = 0; + + if (test_bit(UPROBE_COPY_INSN, &uprobe->flags)) + return ret; + + mutex_lock(&uprobe->copy_mutex); + if (test_bit(UPROBE_COPY_INSN, &uprobe->flags)) + goto out; + + ret = copy_insn(uprobe, file); + if (ret) + goto out; + + ret = -ENOTSUPP; + if (is_swbp_insn((uprobe_opcode_t *)uprobe->arch.insn)) + goto out; + + ret = arch_uprobe_analyze_insn(&uprobe->arch, mm, vaddr); + if (ret) + goto out; + + /* write_opcode() assumes we don't cross page boundary */ + BUG_ON((uprobe->offset & ~PAGE_MASK) + + UPROBE_SWBP_INSN_SIZE > PAGE_SIZE); + + smp_wmb(); /* pairs with rmb() in find_active_uprobe() */ + set_bit(UPROBE_COPY_INSN, &uprobe->flags); + + out: + mutex_unlock(&uprobe->copy_mutex); + + return ret; +} + static int install_breakpoint(struct uprobe *uprobe, struct mm_struct *mm, struct vm_area_struct *vma, unsigned long vaddr) { + bool first_uprobe; int ret; /* @@ -659,48 +625,38 @@ install_breakpoint(struct uprobe *uprobe, struct mm_struct *mm, * Hence behave as if probe already existed. */ if (!uprobe->consumers) - return -EEXIST; - - if (!(uprobe->flags & UPROBE_COPY_INSN)) { - ret = copy_insn(uprobe, vma->vm_file); - if (ret) - return ret; - - if (is_swbp_insn((uprobe_opcode_t *)uprobe->arch.insn)) - return -ENOTSUPP; - - ret = arch_uprobe_analyze_insn(&uprobe->arch, mm, vaddr); - if (ret) - return ret; - - /* write_opcode() assumes we don't cross page boundary */ - BUG_ON((uprobe->offset & ~PAGE_MASK) + - UPROBE_SWBP_INSN_SIZE > PAGE_SIZE); + return 0; - uprobe->flags |= UPROBE_COPY_INSN; - } + ret = prepare_uprobe(uprobe, vma->vm_file, mm, vaddr); + if (ret) + return ret; /* - * Ideally, should be updating the probe count after the breakpoint - * has been successfully inserted. However a thread could hit the - * breakpoint we just inserted even before the probe count is - * incremented. If this is the first breakpoint placed, breakpoint - * notifier might ignore uprobes and pass the trap to the thread. - * Hence increment before and decrement on failure. + * set MMF_HAS_UPROBES in advance for uprobe_pre_sstep_notifier(), + * the task can hit this breakpoint right after __replace_page(). */ - atomic_inc(&mm->uprobes_state.count); + first_uprobe = !test_bit(MMF_HAS_UPROBES, &mm->flags); + if (first_uprobe) + set_bit(MMF_HAS_UPROBES, &mm->flags); + ret = set_swbp(&uprobe->arch, mm, vaddr); - if (ret) - atomic_dec(&mm->uprobes_state.count); + if (!ret) + clear_bit(MMF_RECALC_UPROBES, &mm->flags); + else if (first_uprobe) + clear_bit(MMF_HAS_UPROBES, &mm->flags); return ret; } -static void +static int remove_breakpoint(struct uprobe *uprobe, struct mm_struct *mm, unsigned long vaddr) { - if (!set_orig_insn(&uprobe->arch, mm, vaddr, true)) - atomic_dec(&mm->uprobes_state.count); + /* can happen if uprobe_register() fails */ + if (!test_bit(MMF_HAS_UPROBES, &mm->flags)) + return 0; + + set_bit(MMF_RECALC_UPROBES, &mm->flags); + return set_orig_insn(&uprobe->arch, mm, vaddr); } /* @@ -710,11 +666,9 @@ remove_breakpoint(struct uprobe *uprobe, struct mm_struct *mm, unsigned long vad */ static void delete_uprobe(struct uprobe *uprobe) { - unsigned long flags; - - spin_lock_irqsave(&uprobes_treelock, flags); + spin_lock(&uprobes_treelock); rb_erase(&uprobe->rb_node, &uprobes_tree); - spin_unlock_irqrestore(&uprobes_treelock, flags); + spin_unlock(&uprobes_treelock); iput(uprobe->inode); put_uprobe(uprobe); atomic_dec(&uprobe_events); @@ -818,7 +772,7 @@ static int register_for_each_vma(struct uprobe *uprobe, bool is_register) struct mm_struct *mm = info->mm; struct vm_area_struct *vma; - if (err) + if (err && is_register) goto free; down_write(&mm->mmap_sem); @@ -831,17 +785,11 @@ static int register_for_each_vma(struct uprobe *uprobe, bool is_register) vaddr_to_offset(vma, info->vaddr) != uprobe->offset) goto unlock; - if (is_register) { + if (is_register) err = install_breakpoint(uprobe, mm, vma, info->vaddr); - /* - * We can race against uprobe_mmap(), see the - * comment near uprobe_hash(). - */ - if (err == -EEXIST) - err = 0; - } else { - remove_breakpoint(uprobe, mm, info->vaddr); - } + else + err |= remove_breakpoint(uprobe, mm, info->vaddr); + unlock: up_write(&mm->mmap_sem); free: @@ -897,21 +845,25 @@ int uprobe_register(struct inode *inode, loff_t offset, struct uprobe_consumer * mutex_lock(uprobes_hash(inode)); uprobe = alloc_uprobe(inode, offset); - if (uprobe && !consumer_add(uprobe, uc)) { + if (!uprobe) { + ret = -ENOMEM; + } else if (!consumer_add(uprobe, uc)) { ret = __uprobe_register(uprobe); if (ret) { uprobe->consumers = NULL; __uprobe_unregister(uprobe); } else { - uprobe->flags |= UPROBE_RUN_HANDLER; + set_bit(UPROBE_RUN_HANDLER, &uprobe->flags); } } mutex_unlock(uprobes_hash(inode)); - put_uprobe(uprobe); + if (uprobe) + put_uprobe(uprobe); return ret; } +EXPORT_SYMBOL_GPL(uprobe_register); /* * uprobe_unregister - unregister a already registered probe. @@ -935,7 +887,7 @@ void uprobe_unregister(struct inode *inode, loff_t offset, struct uprobe_consume if (consumer_del(uprobe, uc)) { if (!uprobe->consumers) { __uprobe_unregister(uprobe); - uprobe->flags &= ~UPROBE_RUN_HANDLER; + clear_bit(UPROBE_RUN_HANDLER, &uprobe->flags); } } @@ -943,6 +895,7 @@ void uprobe_unregister(struct inode *inode, loff_t offset, struct uprobe_consume if (uprobe) put_uprobe(uprobe); } +EXPORT_SYMBOL_GPL(uprobe_unregister); static struct rb_node * find_node_in_range(struct inode *inode, loff_t min, loff_t max) @@ -978,7 +931,6 @@ static void build_probe_list(struct inode *inode, struct list_head *head) { loff_t min, max; - unsigned long flags; struct rb_node *n, *t; struct uprobe *u; @@ -986,7 +938,7 @@ static void build_probe_list(struct inode *inode, min = vaddr_to_offset(vma, start); max = min + (end - start) - 1; - spin_lock_irqsave(&uprobes_treelock, flags); + spin_lock(&uprobes_treelock); n = find_node_in_range(inode, min, max); if (n) { for (t = n; t; t = rb_prev(t)) { @@ -1004,27 +956,20 @@ static void build_probe_list(struct inode *inode, atomic_inc(&u->ref); } } - spin_unlock_irqrestore(&uprobes_treelock, flags); + spin_unlock(&uprobes_treelock); } /* - * Called from mmap_region. - * called with mm->mmap_sem acquired. - * - * Return -ve no if we fail to insert probes and we cannot - * bail-out. - * Return 0 otherwise. i.e: + * Called from mmap_region/vma_adjust with mm->mmap_sem acquired. * - * - successful insertion of probes - * - (or) no possible probes to be inserted. - * - (or) insertion of probes failed but we can bail-out. + * Currently we ignore all errors and always return 0, the callers + * can't handle the failure anyway. */ int uprobe_mmap(struct vm_area_struct *vma) { struct list_head tmp_list; struct uprobe *uprobe, *u; struct inode *inode; - int ret, count; if (!atomic_read(&uprobe_events) || !valid_vma(vma, true)) return 0; @@ -1036,44 +981,35 @@ int uprobe_mmap(struct vm_area_struct *vma) mutex_lock(uprobes_mmap_hash(inode)); build_probe_list(inode, vma, vma->vm_start, vma->vm_end, &tmp_list); - ret = 0; - count = 0; - list_for_each_entry_safe(uprobe, u, &tmp_list, pending_list) { - if (!ret) { + if (!fatal_signal_pending(current)) { unsigned long vaddr = offset_to_vaddr(vma, uprobe->offset); - - ret = install_breakpoint(uprobe, vma->vm_mm, vma, vaddr); - /* - * We can race against uprobe_register(), see the - * comment near uprobe_hash(). - */ - if (ret == -EEXIST) { - ret = 0; - - if (!is_swbp_at_addr(vma->vm_mm, vaddr)) - continue; - - /* - * Unable to insert a breakpoint, but - * breakpoint lies underneath. Increment the - * probe count. - */ - atomic_inc(&vma->vm_mm->uprobes_state.count); - } - - if (!ret) - count++; + install_breakpoint(uprobe, vma->vm_mm, vma, vaddr); } put_uprobe(uprobe); } - mutex_unlock(uprobes_mmap_hash(inode)); - if (ret) - atomic_sub(count, &vma->vm_mm->uprobes_state.count); + return 0; +} - return ret; +static bool +vma_has_uprobes(struct vm_area_struct *vma, unsigned long start, unsigned long end) +{ + loff_t min, max; + struct inode *inode; + struct rb_node *n; + + inode = vma->vm_file->f_mapping->host; + + min = vaddr_to_offset(vma, start); + max = min + (end - start) - 1; + + spin_lock(&uprobes_treelock); + n = find_node_in_range(inode, min, max); + spin_unlock(&uprobes_treelock); + + return !!n; } /* @@ -1081,37 +1017,18 @@ int uprobe_mmap(struct vm_area_struct *vma) */ void uprobe_munmap(struct vm_area_struct *vma, unsigned long start, unsigned long end) { - struct list_head tmp_list; - struct uprobe *uprobe, *u; - struct inode *inode; - if (!atomic_read(&uprobe_events) || !valid_vma(vma, false)) return; if (!atomic_read(&vma->vm_mm->mm_users)) /* called by mmput() ? */ return; - if (!atomic_read(&vma->vm_mm->uprobes_state.count)) - return; - - inode = vma->vm_file->f_mapping->host; - if (!inode) + if (!test_bit(MMF_HAS_UPROBES, &vma->vm_mm->flags) || + test_bit(MMF_RECALC_UPROBES, &vma->vm_mm->flags)) return; - mutex_lock(uprobes_mmap_hash(inode)); - build_probe_list(inode, vma, start, end, &tmp_list); - - list_for_each_entry_safe(uprobe, u, &tmp_list, pending_list) { - unsigned long vaddr = offset_to_vaddr(vma, uprobe->offset); - /* - * An unregister could have removed the probe before - * unmap. So check before we decrement the count. - */ - if (is_swbp_at_addr(vma->vm_mm, vaddr) == 1) - atomic_dec(&vma->vm_mm->uprobes_state.count); - put_uprobe(uprobe); - } - mutex_unlock(uprobes_mmap_hash(inode)); + if (vma_has_uprobes(vma, start, end)) + set_bit(MMF_RECALC_UPROBES, &vma->vm_mm->flags); } /* Slot allocation for XOL */ @@ -1213,13 +1130,15 @@ void uprobe_clear_state(struct mm_struct *mm) kfree(area); } -/* - * uprobe_reset_state - Free the area allocated for slots. - */ -void uprobe_reset_state(struct mm_struct *mm) +void uprobe_dup_mmap(struct mm_struct *oldmm, struct mm_struct *newmm) { - mm->uprobes_state.xol_area = NULL; - atomic_set(&mm->uprobes_state.count, 0); + newmm->uprobes_state.xol_area = NULL; + + if (test_bit(MMF_HAS_UPROBES, &oldmm->flags)) { + set_bit(MMF_HAS_UPROBES, &newmm->flags); + /* unconditionally, dup_mmap() skips VM_DONTCOPY vmas */ + set_bit(MMF_RECALC_UPROBES, &newmm->flags); + } } /* @@ -1430,13 +1349,57 @@ bool uprobe_deny_signal(void) */ static bool can_skip_sstep(struct uprobe *uprobe, struct pt_regs *regs) { - if (arch_uprobe_skip_sstep(&uprobe->arch, regs)) - return true; - - uprobe->flags &= ~UPROBE_SKIP_SSTEP; + if (test_bit(UPROBE_SKIP_SSTEP, &uprobe->flags)) { + if (arch_uprobe_skip_sstep(&uprobe->arch, regs)) + return true; + clear_bit(UPROBE_SKIP_SSTEP, &uprobe->flags); + } return false; } +static void mmf_recalc_uprobes(struct mm_struct *mm) +{ + struct vm_area_struct *vma; + + for (vma = mm->mmap; vma; vma = vma->vm_next) { + if (!valid_vma(vma, false)) + continue; + /* + * This is not strictly accurate, we can race with + * uprobe_unregister() and see the already removed + * uprobe if delete_uprobe() was not yet called. + */ + if (vma_has_uprobes(vma, vma->vm_start, vma->vm_end)) + return; + } + + clear_bit(MMF_HAS_UPROBES, &mm->flags); +} + +static int is_swbp_at_addr(struct mm_struct *mm, unsigned long vaddr) +{ + struct page *page; + uprobe_opcode_t opcode; + int result; + + pagefault_disable(); + result = __copy_from_user_inatomic(&opcode, (void __user*)vaddr, + sizeof(opcode)); + pagefault_enable(); + + if (likely(result == 0)) + goto out; + + result = get_user_pages(NULL, mm, vaddr, 1, 0, 1, &page, NULL); + if (result < 0) + return result; + + copy_opcode(page, vaddr, &opcode); + put_page(page); + out: + return is_swbp_insn(&opcode); +} + static struct uprobe *find_active_uprobe(unsigned long bp_vaddr, int *is_swbp) { struct mm_struct *mm = current->mm; @@ -1458,11 +1421,24 @@ static struct uprobe *find_active_uprobe(unsigned long bp_vaddr, int *is_swbp) } else { *is_swbp = -EFAULT; } + + if (!uprobe && test_and_clear_bit(MMF_RECALC_UPROBES, &mm->flags)) + mmf_recalc_uprobes(mm); up_read(&mm->mmap_sem); return uprobe; } +void __weak arch_uprobe_enable_step(struct arch_uprobe *arch) +{ + user_enable_single_step(current); +} + +void __weak arch_uprobe_disable_step(struct arch_uprobe *arch) +{ + user_disable_single_step(current); +} + /* * Run handler and ask thread to singlestep. * Ensure all non-fatal signals cannot interrupt thread while it singlesteps. @@ -1494,41 +1470,42 @@ static void handle_swbp(struct pt_regs *regs) } return; } + /* + * TODO: move copy_insn/etc into _register and remove this hack. + * After we hit the bp, _unregister + _register can install the + * new and not-yet-analyzed uprobe at the same address, restart. + */ + smp_rmb(); /* pairs with wmb() in install_breakpoint() */ + if (unlikely(!test_bit(UPROBE_COPY_INSN, &uprobe->flags))) + goto restart; utask = current->utask; if (!utask) { utask = add_utask(); /* Cannot allocate; re-execute the instruction. */ if (!utask) - goto cleanup_ret; + goto restart; } - utask->active_uprobe = uprobe; + handler_chain(uprobe, regs); - if (uprobe->flags & UPROBE_SKIP_SSTEP && can_skip_sstep(uprobe, regs)) - goto cleanup_ret; + if (can_skip_sstep(uprobe, regs)) + goto out; - utask->state = UTASK_SSTEP; if (!pre_ssout(uprobe, regs, bp_vaddr)) { - user_enable_single_step(current); + arch_uprobe_enable_step(&uprobe->arch); + utask->active_uprobe = uprobe; + utask->state = UTASK_SSTEP; return; } -cleanup_ret: - if (utask) { - utask->active_uprobe = NULL; - utask->state = UTASK_RUNNING; - } - if (uprobe) { - if (!(uprobe->flags & UPROBE_SKIP_SSTEP)) - - /* - * cannot singlestep; cannot skip instruction; - * re-execute the instruction. - */ - instruction_pointer_set(regs, bp_vaddr); - - put_uprobe(uprobe); - } +restart: + /* + * cannot singlestep; cannot skip instruction; + * re-execute the instruction. + */ + instruction_pointer_set(regs, bp_vaddr); +out: + put_uprobe(uprobe); } /* @@ -1547,10 +1524,10 @@ static void handle_singlestep(struct uprobe_task *utask, struct pt_regs *regs) else WARN_ON_ONCE(1); + arch_uprobe_disable_step(&uprobe->arch); put_uprobe(uprobe); utask->active_uprobe = NULL; utask->state = UTASK_RUNNING; - user_disable_single_step(current); xol_free_insn_slot(current); spin_lock_irq(&current->sighand->siglock); @@ -1559,13 +1536,12 @@ static void handle_singlestep(struct uprobe_task *utask, struct pt_regs *regs) } /* - * On breakpoint hit, breakpoint notifier sets the TIF_UPROBE flag. (and on - * subsequent probe hits on the thread sets the state to UTASK_BP_HIT) and - * allows the thread to return from interrupt. + * On breakpoint hit, breakpoint notifier sets the TIF_UPROBE flag and + * allows the thread to return from interrupt. After that handle_swbp() + * sets utask->active_uprobe. * - * On singlestep exception, singlestep notifier sets the TIF_UPROBE flag and - * also sets the state to UTASK_SSTEP_ACK and allows the thread to return from - * interrupt. + * On singlestep exception, singlestep notifier sets the TIF_UPROBE flag + * and allows the thread to return from interrupt. * * While returning to userspace, thread notices the TIF_UPROBE flag and calls * uprobe_notify_resume(). @@ -1574,11 +1550,13 @@ void uprobe_notify_resume(struct pt_regs *regs) { struct uprobe_task *utask; + clear_thread_flag(TIF_UPROBE); + utask = current->utask; - if (!utask || utask->state == UTASK_BP_HIT) - handle_swbp(regs); - else + if (utask && utask->active_uprobe) handle_singlestep(utask, regs); + else + handle_swbp(regs); } /* @@ -1587,18 +1565,10 @@ void uprobe_notify_resume(struct pt_regs *regs) */ int uprobe_pre_sstep_notifier(struct pt_regs *regs) { - struct uprobe_task *utask; - - if (!current->mm || !atomic_read(&current->mm->uprobes_state.count)) - /* task is currently not uprobed */ + if (!current->mm || !test_bit(MMF_HAS_UPROBES, &current->mm->flags)) return 0; - utask = current->utask; - if (utask) - utask->state = UTASK_BP_HIT; - set_thread_flag(TIF_UPROBE); - return 1; } diff --git a/kernel/fork.c b/kernel/fork.c index 2c8857e..2343c9e 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -353,6 +353,7 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) down_write(&oldmm->mmap_sem); flush_cache_dup_mm(oldmm); + uprobe_dup_mmap(oldmm, mm); /* * Not linked in yet - no deadlock potential: */ @@ -454,9 +455,6 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) if (retval) goto out; - - if (file) - uprobe_mmap(tmp); } /* a new mm has just been created */ arch_dup_mmap(oldmm, mm); @@ -839,8 +837,6 @@ struct mm_struct *dup_mm(struct task_struct *tsk) #ifdef CONFIG_TRANSPARENT_HUGEPAGE mm->pmd_huge_pte = NULL; #endif - uprobe_reset_state(mm); - if (!mm_init(mm, tsk)) goto fail_nomem; diff --git a/kernel/ptrace.c b/kernel/ptrace.c index a232bb5..764fcd1 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -33,6 +33,12 @@ static int ptrace_trapping_sleep_fn(void *flags) } /* + * This is declared in linux/regset.h and defined in machine-dependent + * code. We put the export here to ensure no machine forgets it. + */ +EXPORT_SYMBOL_GPL(task_user_regset_view); + +/* * ptrace a task: make the debugger its new parent and * move it to the ptrace list. *

2 1

Re: FUDCon Paris kernel talk
by Thorsten Leemhuis 26 Oct '12

26 Oct '12

Hi! Jwb in http://jwboyer.livejournal.com/44908.html wrote: > Now I have a question for you, whether you were at this talk, have > seen a similar one at a different FUDCon, or want to speak up in > general: Are these talks worthwhile to you? I haven't been to Paris and just listened to two (or was it just one? can't remember) Fudcon kernel talks from davej years ago. I found them useful, as they gave some new insights and I learned a thing or two about how the Fedora's kernel crew worked back then. But the real reason why I write this mail: I like to thank Fedora's kernel team and especially Josh for blog entries like the one mentioned above. Those are IMHO more important than FUDCon talks, as everybody interested can get information this way -- at a time they want and without travelling long distances. Further: a text like this can transfer more information in less time than a talk or a video recording from a talk (which in the FUDCon Paris case afaik doesn't even exist); and in a text it's a lot easier to skip details one is not interested in. In addition: Thanks for the public IRC meetings and the summaries from them. They give a good insight, too. I hope this mail encourages you guys to keep on writing blog posts and meetings summaries (or maybe even do it more often). I know the feedback and responses to them are rare, which can be frustrating and easily lead to a "nobody seems to read that stuff, so why do I bother and spend time on it" attitude (trust me, I know what I'm talking about ;-) ). But be assured: I read them, and I'm sure many others do, too; so keep up with it, it helps a lot! :-) Cu knurd

3 2

An RFC related to Fedora Kernel Testing Initiative
by Steffen Schwigon 23 Oct '12

23 Oct '12

Hi! Several weeks ago I read about https://fedoraproject.org/wiki/KernelTestingInitiative https://fedoraproject.org/wiki/KernelRegressionTestGuidelines and recognized the vision is very similar to what we exercise for several years in the AMD Operating System Research Center when writing functional tests for Linux, Xen, and KVM. After some polishing I just submitted an RFC: A common test container https://lkml.org/lkml/2012/10/23/101 https://lkml.org/lkml/2012/10/23/105 https://lkml.org/lkml/2012/10/23/102 to LKML. I explicitely had the Fedora Kernel Testing Initiative in mind, I even integrated a last missing part from the fedora testing vision into the proposed lib: the ability to skip tests based on their "criticality level". But from several years of experience with Linux testing I think the best place is to put a common stub *in* the kernel, and be able to proceed from there. If you are interested feel free to join the discussion on LKML, or if you prefer tell me to forward it here. Kind regards, Steffen -- Steffen Schwigon <steffen.schwigon(a)amd.com> AMD, Operating System Research Center

1 0

[PATCH F18] xen/pv-on-hvm kexec: add quirk for Xen 3.4 and shutdown watches
by Miroslav Rezanina 22 Oct '12

22 Oct '12

This is backport of upstream commit cb6b6df111e46b9d0f79eb971575fd50555f43f4 that is required to run Fedora 18 as HVM guest on RHEL 5 Xen hypervisor. xen/pv-on-hvm kexec: add quirk for Xen 3.4 and shutdown watches. The commit 254d1a3f02ebc10ccc6e4903394d8d3f484f715e, titled "xen/pv-on-hvm kexec: shutdown watches from old kernel" assumes that the XenBus backend can deal with reading of values from: "control/platform-feature-xs_reset_watches": ... a patch for xenstored is required so that it accepts the XS_RESET_WATCHES request from a client (see changeset 23839:42a45baf037d in xen-unstable.hg). Without the patch for xenstored the registration of watches will fail and some features of a PVonHVM guest are not available. The guest is still able to boot, but repeated kexec boots will fail." Sadly this is not true when using a Xen 3.4 hypervisor and booting a PVHVM guest. We end up hanging at: err = xenbus_scanf(XBT_NIL, "control", "platform-feature-xs_reset_watches", "%d", &supported); This can easily be seen with guests hanging at xenbus_init: NX (Execute Disable) protection: active SMBIOS 2.4 present. DMI: Xen HVM domU, BIOS 3.4.0 05/13/2011 Hypervisor detected: Xen HVM Xen version 3.4. Xen Platform PCI: I/O protocol version 1 ... snip .. calling xenbus_init+0x0/0x27e @ 1 Reverting the commit or using the attached patch fixes the issue. This fix checks whether the hypervisor is older than 4.0 and if so does not try to perform the read. Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk(a)oracle.com> Signed-off-by: Miroslav Rezanina <mrezanin(a)redhat.com> diff --git a/drivers/xen/xenbus/xenbus_xs.c b/drivers/xen/xenbus/xenbus_xs.c index bce15cf..ca373d1 100644 --- a/drivers/xen/xenbus/xenbus_xs.c +++ b/drivers/xen/xenbus/xenbus_xs.c @@ -47,6 +47,7 @@ #include <xen/xenbus.h> #include <xen/xen.h> #include "xenbus_comms.h" +#include <asm/xen/hypervisor.h> struct xs_stored_msg { struct list_head list; @@ -617,7 +618,24 @@ static struct xenbus_watch *find_watch(const char *token) return NULL; } +/* + * Certain older XenBus toolstack cannot handle reading values that are + * not populated. Some Xen 3.4 installation are incapable of doing this + * so if we are running on anything older than 4 do not attempt to read + * control/platform-feature-xs_reset_watches. + */ +static bool xen_strict_xenbus_quirk() +{ + uint32_t eax, ebx, ecx, edx, base; + + base = xen_cpuid_base(); + cpuid(base + 1, &eax, &ebx, &ecx, &edx); + if ((eax >> 16) < 4) + return true; + return false; + +} static void xs_reset_watches(void) { int err, supported = 0; @@ -625,6 +643,9 @@ static void xs_reset_watches(void) if (!xen_hvm_domain()) return; + if (xen_strict_xenbus_quirk()) + return; + err = xenbus_scanf(XBT_NIL, "control", "platform-feature-xs_reset_watches", "%d", &supported); if (err != 1 || !supported)

2 2

pesign
by Mr Dash Four 21 Oct '12

21 Oct '12

I seem to remember in one of the early 3.6-RC kernel versions there were provisions put in the .spec file to sign all kernel code and its modules using the above facility. I can't find this in the 3.6.1 or 3.6.2 versions of the kernel currently in the Fedora srpm files. Has this been dropped? On a related issue - if, for some reason, I am unable to deploy UEFI (disabled, so that Windows 8 won't prevent me from installing/using/booting up Linux) can I still sign the kernel and its modules and enforce these checks at startup with the bootloader (grub2)? Would that be possible? Thanks!

2 5

Patch: move power-x86-destdir.patch to a section where it is applied even for vanilla builds
by Thorsten Leemhuis 20 Oct '12

20 Oct '12

Hi! power-x86-destdir.patch should be applied always, as building the Fedora spec file with the switch "--with vanilla" will fail otherwise. Attached patch fixes that for rawhide. BTW: is that patch heading upstream? While at it: Is it really a good idea to put the definitions for Fedora specific patches in a section with a "%if !%{nopatches}" conditional? If someone grabs the SRPM from a vanilla build then he can't properly recompile it without using "--with vanilla", as the SRPM misses the patches for a proper Fedora build. Sure, that's a corner case, but the space saving from omitting the patches are not that big, so I was wondering if this "%if !%{nopatches}" conditional is better removed, as it's not worth the trouble. Cu knurd

1 1

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

kernel October 2012