kthread detection cleanup
Subsystems (46)
Mark Rutland
maintainercover kthread detection cleanup
A reasonable amount of kernel code looks at task_struct::mm to determine whether a thread is a kthread or a real user task. This isn't quite right, since kthreads can have a non-NULL mm when calling use_mm().
The correct way to check whether a task is a kthread is to check whether PF_KTHREAD is set in task_struct::flags, but doing so is a bit unwieldy.
To make this a bit nicer, this series adds a new is_kthread(tsk) helper, converts existing code to make use of it, and fixes up a number of erroneous checks of current->mm. Hopefully this will push people to use the right approach in future.
I'm sure there are other instances in the kernel tree where we don't check this correctly. In this series I'm just trying to fix the instances I'm reasonably confident are incorrect.
This series is based on v5.3-rc3, and I've pushed it out to my sched/kthread-cleanup branch on kernel.org [1,2].
Thanks,
Mark.
[1] git://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git sched/kthread-cleanup
[2] https://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git/log/?h=sched/kthread-cleanup
Mark Rutland (9):
sched/core: add is_kthread() helper
sched: treewide: use is_kthread()
kmemleak: correctly check for kthreads
arm/perf: correctly check for kthreads
arm64: correctly check for ktrheads
sparc/perf: correctly check for kthreads
x86/lbr: correctly check for kthreads
x86/fpu: correctly check for kthreads
samples/kretprobe: correctly check for kthreads
arch/alpha/kernel/process.c | 2 +- arch/arc/kernel/process.c | 2 +- arch/arm/kernel/perf_callchain.c | 2 +- arch/arm/kernel/process.c | 2 +- arch/arm/mm/init.c | 2 +- arch/arm64/kernel/process.c | 6 +++--- arch/c6x/kernel/process.c | 2 +- arch/csky/kernel/process.c | 2 +- arch/h8300/kernel/process.c | 2 +- arch/hexagon/kernel/process.c | 2 +- arch/ia64/kernel/process.c | 2 +- arch/m68k/kernel/process.c | 2 +- arch/microblaze/kernel/process.c | 2 +- arch/mips/kernel/process.c | 4 ++-- arch/nds32/kernel/process.c | 4 ++-- arch/nios2/kernel/process.c | 2 +- arch/openrisc/kernel/process.c | 2 +- arch/parisc/kernel/process.c | 2 +- arch/powerpc/kernel/process.c | 2 +- arch/riscv/kernel/process.c | 2 +- arch/s390/kernel/process.c | 2 +- arch/sh/kernel/process_32.c | 2 +- arch/sh/kernel/process_64.c | 2 +- arch/sparc/kernel/perf_event.c | 2 +- arch/sparc/kernel/process_32.c | 2 +- arch/sparc/kernel/process_64.c | 2 +- arch/um/kernel/process.c | 2 +- arch/unicore32/kernel/process.c | 2 +- arch/x86/events/intel/lbr.c | 2 +- arch/x86/include/asm/fpu/internal.h | 2 +- arch/x86/kernel/fpu/core.c | 2 +- arch/x86/kernel/process_32.c | 2 +- arch/x86/kernel/process_64.c | 2 +- arch/xtensa/kernel/process.c | 2 +- block/blk-cgroup.c | 2 +- drivers/tty/sysrq.c | 2 +- fs/coredump.c | 2 +- fs/file_table.c | 4 ++-- fs/namespace.c | 2 +- fs/proc/base.c | 4 ++-- include/linux/cgroup.h | 2 +- include/linux/sched.h | 13 ++++++++++++- kernel/cgroup/freezer.c | 4 ++-- kernel/events/core.c | 2 +- kernel/exit.c | 2 +- kernel/fork.c | 6 +++--- kernel/freezer.c | 4 ++-- kernel/futex.c | 2 +- kernel/kthread.c | 6 +++--- kernel/livepatch/transition.c | 2 +- kernel/ptrace.c | 2 +- kernel/sched/core.c | 8 ++++---- kernel/sched/idle.c | 2 +- kernel/sched/wait.c | 2 +- kernel/signal.c | 2 +- kernel/stacktrace.c | 2 +- lib/is_single_threaded.c | 2 +- mm/kmemleak.c | 6 +++--- mm/memcontrol.c | 2 +- mm/oom_kill.c | 4 ++-- mm/page_alloc.c | 2 +- mm/vmacache.c | 2 +- mm/vmscan.c | 2 +- samples/kprobes/kretprobe_example.c | 2 +- security/smack/smack_access.c | 2 +- security/smack/smack_lsm.c | 4 ++-- security/yama/yama_lsm.c | 2 +- 67 files changed, 97 insertions(+), 86 deletions(-)
Signature · 1 line
2.11.0
Mark Rutland
maintainerPATCH 1/9 sched/core: add is_kthread() helper
Code checking whether a task is a kthread isn't very consistent. Some code correctly tests task->flags & PF_THREAD, while other code checks task->mm (which can be true for a kthread which calls use_mm()).
So that we can clean this up and keep the code easy to follow, let's add an obvious helper function to test whether a task is a kthread. Subsequent patches will use this as part of cleaning up and correcting open-coded tests.
At the same time, let's fix up the kerneldoc for is_idle_task() for consistency with the new helper, using true/false rather than 0/1, given the functions return bool.
--- include/linux/sched.h | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-)
include/linux/sched.h +12 -1
@@ -1621,13 +1621,24 @@ extern struct task_struct *idle_task(int cpu);* is_idle_task - is the specified task an idle task? * @p: the task in question. * -* Return: 1 if @p is an idle task. 0 otherwise. +* Return: true if @p is an idle task, false otherwise. */ static inline bool is_idle_task (const struct task_struct *p) {return !!(p->flags &PF_IDLE ); } +/** +* is_kthread - is the specified task a kthread +* @p: the task in question. +* +* Return: true if @p is a kthread, false otherwise. +*/ +static inline bool is_kthread (const struct task_struct *p) +{ +return !!(p->flags &PF_KTHREAD ); +} +extern struct task_struct *curr_task (int cpu);extern void ia64_set_curr_task (int cpu,struct task_struct *p);
Signature · 1 line
2.11.0
Geert Uytterhoeven
maintainerRe: [PATCH 1/9] sched/core: add is_kthread() helper
Hi Mark,
Quoted 9 lines · Mark Rutland
Code checking whether a task is a kthread isn't very consistent. Some code correctly tests
task->flags&PF_THREAD, while other code checkstask->mm(which can be true for a kthread which callsuse_mm()).So that we can clean this up and keep the code easy to follow, let's add an obvious helper function to test whether a task is a kthread. Subsequent patches will use this as part of cleaning up and correcting open-coded tests.
Quoted 5 lines
At the same time, let's fix up the kerneldoc for
is_idle_task()for consistency with the new helper, using true/false rather than 0/1, given the functions return bool.Signed-off-by Mark Rutland <[email protected]>
Thanks for your patch!
Quoted 13 lines
include/linux/sched.h +1
-1@@ -1621,13 +1621,24 @@ extern struct task_struct *idle_task(int cpu);* is_idle_task - is the specified task an idle task? * @p: the task in question. * -* Return: 1 if @p is an idle task. 0 otherwise. +* Return: true if @p is an idle task, false otherwise. */ static inline bool is_idle_task (const struct task_struct *p) {return !!(p->flags &PF_IDLE ); }
include/linux/sched.h extern struct task_struct *idle_task(int cpu); · line 1639
Quoted 10 lines
include/linux/sched.h +9
-0+/** +* is_kthread - is the specified task a kthread +* @p: the task in question. +* +* Return: true if @p is a kthread, false otherwise. +*/ +static inline bool is_kthread (const struct task_struct *p) +{ +return !!(p->flags &PF_KTHREAD );
The !! is not really needed.
Probably you followed is_idle_task() above (where it's also not needed).
include/linux/sched.h extern struct task_struct *idle_task(int cpu); · line 1641
Quoted 2 lines
include/linux/sched.h +2
-0+} +
Gr{oetje,eeting}s,
Geert
Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected] In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds
Mark Rutland
maintainerRe: [PATCH 1/9] sched/core: add is_kthread() helper
Quoted 34 lines · Geert Uytterhoeven
Hi Mark,
Quoted 9 lines · Mark Rutland
Code checking whether a task is a kthread isn't very consistent. Some code correctly tests
task->flags&PF_THREAD, while other code checkstask->mm(which can be true for a kthread which callsuse_mm()).So that we can clean this up and keep the code easy to follow, let's add an obvious helper function to test whether a task is a kthread. Subsequent patches will use this as part of cleaning up and correcting open-coded tests.
Quoted 5 lines
At the same time, let's fix up the kerneldoc for
is_idle_task()for consistency with the new helper, using true/false rather than 0/1, given the functions return bool.Signed-off-by Mark Rutland <[email protected]>Thanks for your patch!
Quoted 14 lines
include/linux/sched.h +1
-1@@ -1621,13 +1621,24 @@ extern struct task_struct *idle_task(int cpu);* is_idle_task - is the specified task an idle task? * @p: the task in question. * -* Return: 1 if @p is an idle task. 0 otherwise. +* Return: true if @p is an idle task, false otherwise. */ static inline bool is_idle_task (const struct task_struct *p) {return !!(p->flags &PF_IDLE ); }
Quoted 12 lines
include/linux/sched.h extern struct task_struct *idle_task(int cpu); · line 1639
Quoted 9 lines
include/linux/sched.h +9
-0+/** +* is_kthread - is the specified task a kthread +* @p: the task in question. +* +* Return: true if @p is a kthread, false otherwise. +*/ +static inline bool is_kthread (const struct task_struct *p) +{ +return !!(p->flags &PF_KTHREAD );The !! is not really needed.
Probably you followedis_idle_task()above (where it's also not needed).
Indeed! I'm aware of the implicit bool conversion, but kept that for consistency.
Peter, Ingo, do you have a preference?
Thanks,
Mark.
Re: [PATCH 1/9] sched/core: add is_kthread() helper
Quoted 48 lines · Mark Rutland
Quoted 34 lines · Geert Uytterhoeven
Hi Mark,
Quoted 9 lines · Mark Rutland
Code checking whether a task is a kthread isn't very consistent. Some code correctly tests
task->flags&PF_THREAD, while other code checkstask->mm(which can be true for a kthread which callsuse_mm()).So that we can clean this up and keep the code easy to follow, let's add an obvious helper function to test whether a task is a kthread. Subsequent patches will use this as part of cleaning up and correcting open-coded tests.
Quoted 5 lines
At the same time, let's fix up the kerneldoc for
is_idle_task()for consistency with the new helper, using true/false rather than 0/1, given the functions return bool.Signed-off-by Mark Rutland <[email protected]>Thanks for your patch!
Quoted 14 lines
include/linux/sched.h +1
-1@@ -1621,13 +1621,24 @@ extern struct task_struct *idle_task(int cpu);* is_idle_task - is the specified task an idle task? * @p: the task in question. * -* Return: 1 if @p is an idle task. 0 otherwise. +* Return: true if @p is an idle task, false otherwise. */ static inline bool is_idle_task (const struct task_struct *p) {return !!(p->flags &PF_IDLE ); }Quoted 12 lines
include/linux/sched.h extern struct task_struct *idle_task(int cpu); · line 1639
Quoted 9 lines
include/linux/sched.h +9
-0+/** +* is_kthread - is the specified task a kthread +* @p: the task in question. +* +* Return: true if @p is a kthread, false otherwise. +*/ +static inline bool is_kthread (const struct task_struct *p) +{ +return !!(p->flags &PF_KTHREAD );The !! is not really needed.
Probably you followedis_idle_task()above (where it's also not needed).
Quoted 4 lines
Indeed! I'm aware of the implicit bool conversion, but kept that for consistency.
Peter, Ingo, do you have a preference?
So the !! pattern is useful where the return value is an integer (i.e. there's a risk of non-bool use) - but the return value is an explicit bool here, so !! is IMO an entirely superfluous obfuscation.
Should probably be fixed for is_idle_task() as well?
Thanks,
Ingo
Peter Zijlstra
maintainerRe: [PATCH 1/9] sched/core: add is_kthread() helper
Quoted 11 lines · Ingo Molnar
Quoted 10 lines · Mark Rutland
Quoted 8 lines · Geert Uytterhoeven
include/linux/sched.h extern struct task_struct *idle_task(int cpu); · line 1639
Quoted 3 lines · Mark Rutland
include/linux/sched.h +3
-0+static inline bool is_kthread (const struct task_struct *p) +{ +return !!(p->flags &PF_KTHREAD );The !! is not really needed.
Probably you followedis_idle_task()above (where it's also not needed).
Quoted 8 lines
Quoted 4 lines
Indeed! I'm aware of the implicit bool conversion, but kept that for consistency.
Peter, Ingo, do you have a preference?
So the !! pattern is useful where the return value is an integer (i.e. there's a risk of non-bool use) - but the return value is an explicit bool here, so !! is IMO an entirely superfluous obfuscation.
Yeah, no real preference, for giggles, (_Bool) also seems to work.
Mark Rutland
maintainerPATCH 2/9 sched: treewide: use is_kthread()
In reply to Mark RutlandNow that we have is_kthread(), let's convert existing open-coded checks of the form:
task->flags & PF_KTHREAD
... over to the new helper, which makes things a little easier to read, and sets a consistent example for new code to follow.
Generated with coccinelle:
----
virtual patch
@ depends on patch @
expression E;
@@
- (E->flags & PF_KTHREAD)
+ is_kthread(E)
----
... though this didn't pick up the instance in <linux/cgroup.h>, which I fixed up manually.
Instances checking multiple PF_* flags at ocne are left as-is for now.
There should be no functional change as a result of this patch.
--- arch/alpha/kernel/process.c | 2 +- arch/arc/kernel/process.c | 2 +- arch/arm/kernel/process.c | 2 +- arch/arm/mm/init.c | 2 +- arch/arm64/kernel/process.c | 4 ++-- arch/c6x/kernel/process.c | 2 +- arch/csky/kernel/process.c | 2 +- arch/h8300/kernel/process.c | 2 +- arch/hexagon/kernel/process.c | 2 +- arch/ia64/kernel/process.c | 2 +- arch/m68k/kernel/process.c | 2 +- arch/microblaze/kernel/process.c | 2 +- arch/mips/kernel/process.c | 4 ++-- arch/nds32/kernel/process.c | 4 ++-- arch/nios2/kernel/process.c | 2 +- arch/openrisc/kernel/process.c | 2 +- arch/parisc/kernel/process.c | 2 +- arch/powerpc/kernel/process.c | 2 +- arch/riscv/kernel/process.c | 2 +- arch/s390/kernel/process.c | 2 +- arch/sh/kernel/process_32.c | 2 +- arch/sh/kernel/process_64.c | 2 +- arch/sparc/kernel/process_32.c | 2 +- arch/sparc/kernel/process_64.c | 2 +- arch/um/kernel/process.c | 2 +- arch/unicore32/kernel/process.c | 2 +- arch/x86/kernel/fpu/core.c | 2 +- arch/x86/kernel/process_32.c | 2 +- arch/x86/kernel/process_64.c | 2 +- arch/xtensa/kernel/process.c | 2 +- block/blk-cgroup.c | 2 +- drivers/tty/sysrq.c | 2 +- fs/coredump.c | 2 +- fs/file_table.c | 4 ++-- fs/namespace.c | 2 +- fs/proc/base.c | 4 ++-- include/linux/cgroup.h | 2 +- kernel/cgroup/freezer.c | 4 ++-- kernel/events/core.c | 2 +- kernel/exit.c | 2 +- kernel/fork.c | 6 +++--- kernel/freezer.c | 4 ++-- kernel/futex.c | 2 +- kernel/kthread.c | 6 +++--- kernel/livepatch/transition.c | 2 +- kernel/ptrace.c | 2 +- kernel/sched/core.c | 8 ++++---- kernel/sched/idle.c | 2 +- kernel/sched/wait.c | 2 +- kernel/signal.c | 2 +- kernel/stacktrace.c | 2 +- lib/is_single_threaded.c | 2 +- mm/memcontrol.c | 2 +- mm/oom_kill.c | 4 ++-- mm/page_alloc.c | 2 +- mm/vmacache.c | 2 +- mm/vmscan.c | 2 +- security/smack/smack_access.c | 2 +- security/smack/smack_lsm.c | 4 ++-- security/yama/yama_lsm.c | 2 +- 60 files changed, 76 insertions(+), 76 deletions(-)
arch/alpha/kernel/process.c +1 -1
@@ -251,7 +251,7 @@ copy_thread(unsigned long clone_flags, unsigned long usp, childti->pcb.ksp = (unsigned long ) childstack; childti->pcb.flags =1 ;/* set FEN, clear everything else */ -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (childstack,0 ,sizeof (struct switch_stack ) +sizeof (struct pt_regs ));
arch/arc/kernel/process.c +1 -1
@@ -200,7 +200,7 @@ int copy_thread(unsigned long clone_flags, childksp[0 ] =0 ;/* fp */ childksp[1 ] = (unsigned long )ret_from_fork;/* blink */ -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (c_regs,0 ,sizeof (struct pt_regs )); c_callee->r13 = kthread_arg;
arch/arm/kernel/process.c +1 -1
@@ -242,7 +242,7 @@ copy_thread(unsigned long clone_flags, unsigned long stack_start, thread->cpu_domain =get_domain ();#endif -if (likely (!(p->flags &PF_KTHREAD ))) { +if (likely (!is_kthread (p))) { *childregs = *current_pt_regs (); childregs->ARM_r0 =0 ;if (stack_start)
arch/arm/mm/init.c +1 -1
@@ -625,7 +625,7 @@ static void update_sections_early(struct section_perm perms[], int n)struct task_struct *t, *s;for_each_process (t) { -if (t->flags &PF_KTHREAD ) +if (is_kthread (t))continue ;for_each_thread (t, s)set_section_perms (perms, n, true, s->mm);
arch/arm64/kernel/process.c +2 -2
@@ -367,7 +367,7 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,*/ fpsimd_flush_task_state (p); -if (likely (!(p->flags &PF_KTHREAD ))) { +if (likely (!is_kthread (p))) { *childregs = *current_pt_regs (); childregs->regs[0 ] =0 ;
@@ -454,7 +454,7 @@ static void ssbs_thread_switch(struct task_struct *next)* Nothing to do for kernel threads, but 'regs' may be junk * (e.g. idle task) so check the flags and bail early. */ -if (unlikely (next->flags &PF_KTHREAD )) +if (unlikely (is_kthread (next)))return ;/* If the mitigation is enabled, then we leave SSBS clear. */
arch/c6x/kernel/process.c +1 -1
@@ -112,7 +112,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, childregs =task_pt_regs (p); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* case of __kernel_thread: we return to supervisor space */ memset (childregs,0 ,sizeof (struct pt_regs )); childregs->sp = (unsigned long )(childregs +1 );
arch/csky/kernel/process.c +1 -1
@@ -52,7 +52,7 @@ int copy_thread(unsigned long clone_flags,/* setup ksp for switch_to !!! */ p->thread.ksp = (unsigned long )childstack; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childregs,0 ,sizeof (struct pt_regs )); childstack->r15 = (unsigned long ) ret_from_kernel_thread; childstack->r8 = kthread_arg;
arch/h8300/kernel/process.c +1 -1
@@ -114,7 +114,7 @@ int copy_thread(unsigned long clone_flags, childregs = (struct pt_regs *) (THREAD_SIZE +task_stack_page (p)) -1 ; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childregs,0 ,sizeof (struct pt_regs )); childregs->retpc = (unsigned long ) ret_from_kernel_thread; childregs->er4 = topstk;/* arg */
arch/hexagon/kernel/process.c +1 -1
@@ -73,7 +73,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, sizeof(*ss)); ss->lr = (unsigned long )ret_from_fork; p->thread.switch_sp = ss; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childregs,0 ,sizeof (struct pt_regs ));/* r24 <- fn, r25 <- arg */ ss->r24 = usp;
arch/ia64/kernel/process.c +1 -1
@@ -376,7 +376,7 @@ copy_thread(unsigned long clone_flags,ia64_drop_fpu (p);/* don't pick up stale state from a CPU's fph */ -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {if (unlikely (!user_stack_base)) {/* fork_idle() called us */ return 0 ;
arch/m68k/kernel/process.c +1 -1
@@ -138,7 +138,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,*/ p->thread.fs =get_fs ().seg; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (frame,0 ,sizeof (struct fork_frame )); frame->regs.sr =PS_S ;
arch/microblaze/kernel/process.c +1 -1
@@ -60,7 +60,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,struct pt_regs *childregs =task_pt_regs (p);struct thread_info *ti =task_thread_info (p); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* if we're creating a new kernel thread then just zeroing all * the registers. That's OK for a brand new thread.*/ memset (childregs,0 ,sizeof (struct pt_regs ));
arch/mips/kernel/process.c +2 -2
@@ -87,7 +87,7 @@ void exit_thread(struct task_struct *tsk)* User threads may have allocated a delay slot emulation frame. * If so, clean up that allocation. */ -if (!(current->flags &PF_KTHREAD )) +if (!is_kthread (current))dsemul_thread_cleanup (tsk); }
@@ -132,7 +132,7 @@ int copy_thread_tls(unsigned long clone_flags, unsigned long usp,/* Put the stack after the struct pt_regs. */ childksp = (unsigned long ) childregs; p->thread.cp0_status =read_c0_status () & ~(ST0_CU2 |ST0_CU1 ); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ unsigned long status = p->thread.cp0_status;memset (childregs,0 ,sizeof (struct pt_regs ));
arch/nds32/kernel/process.c +2 -2
@@ -156,7 +156,7 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,memset (&p->thread.cpu_context,0 ,sizeof (struct cpu_context )); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childregs,0 ,sizeof (struct pt_regs ));/* kernel thread fn */ p->thread.cpu_context.r6 = stack_start;
@@ -207,7 +207,7 @@ struct task_struct *_switch_fpu(struct task_struct *prev, struct task_struct *ne#if !IS_ENABLED (CONFIG_LAZY_FPU )unlazy_fpu (prev);#endif -if (!(next->flags &PF_KTHREAD )) +if (!is_kthread (next))clear_fpu (task_pt_regs (next));return prev; }
arch/nios2/kernel/process.c +1 -1
@@ -109,7 +109,7 @@ int copy_thread(unsigned long clone_flags,struct switch_stack *childstack = ((struct switch_stack *)childregs) -1 ; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childstack,0 ,sizeof (struct switch_stack ) +sizeof (struct pt_regs ));
arch/openrisc/kernel/process.c +1 -1
@@ -168,7 +168,7 @@ copy_thread(unsigned long clone_flags, unsigned long usp, sp -=sizeof (struct pt_regs ); kregs = (struct pt_regs *)sp; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (kregs,0 ,sizeof (struct pt_regs )); kregs->gpr[20 ] = usp;/* fn, kernel thread */ kregs->gpr[22 ] = arg;
arch/parisc/kernel/process.c +1 -1
@@ -220,7 +220,7 @@ copy_thread(unsigned long clone_flags, unsigned long usp,extern void *const ret_from_kernel_thread;extern void *const child_return; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (cregs,0 ,sizeof (struct pt_regs ));if (!usp)/* idle thread */
arch/powerpc/kernel/process.c +1 -1
@@ -1615,7 +1615,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,/* Copy registers */ sp -=sizeof (struct pt_regs ); childregs = (struct pt_regs *) sp; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (childregs,0 ,sizeof (struct pt_regs )); childregs->gpr[1 ] = sp +sizeof (struct pt_regs );
arch/riscv/kernel/process.c +1 -1
@@ -96,7 +96,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,struct pt_regs *childregs =task_pt_regs (p);/* p->thread holds context to be restored by __switch_to() */ -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* Kernel thread */ const registerunsigned long gp __asm__ ("gp" );memset (childregs,0 ,sizeof (struct pt_regs ));
arch/s390/kernel/process.c +1 -1
@@ -113,7 +113,7 @@ int copy_thread_tls(unsigned long clone_flags, unsigned long new_stackp, frame->sf.gprs[9 ] = (unsigned long ) frame;/* Store access registers to kernel stack of new process. */ -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (&frame->childregs,0 ,sizeof (struct pt_regs )); frame->childregs.psw.mask = PSW_KERNEL_BITS | PSW_MASK_DAT |
arch/sh/kernel/process_32.c +1 -1
@@ -137,7 +137,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, childregs =task_pt_regs (p); p->thread.sp = (unsigned long ) childregs; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childregs,0 ,sizeof (struct pt_regs )); p->thread.pc = (unsigned long ) ret_from_kernel_thread; childregs->regs[4 ] = arg;
arch/sh/kernel/process_64.c +1 -1
@@ -389,7 +389,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, childregs = (struct pt_regs *)(THREAD_SIZE +task_stack_page (p)) -1 ; p->thread.sp = (unsigned long ) childregs; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (childregs,0 ,sizeof (struct pt_regs )); childregs->regs[2 ] = (unsigned long )arg; childregs->regs[3 ] = (unsigned long )usp;
arch/sparc/kernel/process_32.c +1 -1
@@ -338,7 +338,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, ti->ksp = (unsigned long ) new_stack; p->thread.kregs = childregs; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {extern int nwindows;unsigned long psr;memset (new_stack,0 ,STACKFRAME_SZ +TRACEREG_SZ );
arch/sparc/kernel/process_64.c +1 -1
@@ -632,7 +632,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, sizeof(struct sparc_stackf)); t->fpsaved[0 ] =0 ; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {memset (child_trap_frame,0 , child_stack_sz);__thread_flag_byte_ptr (t)[TI_FLAG_BYTE_CWP ] = (current_pt_regs ()->tstate +1 ) &TSTATE_CWP ;
arch/um/kernel/process.c +1 -1
@@ -157,7 +157,7 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, unsigned long arg, struct task_struct * p) {void (*handler)(void ); -int kthread = current->flags &PF_KTHREAD ; +int kthread =is_kthread (current);int ret =0 ; p->thread = (struct thread_struct )INIT_THREAD ;
arch/unicore32/kernel/process.c +1 -1
@@ -228,7 +228,7 @@ copy_thread(unsigned long clone_flags, unsigned long stack_start,memset (&thread->cpu_context,0 ,sizeof (struct cpu_context_save )); thread->cpu_context.sp = (unsigned long )childregs; -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) { thread->cpu_context.pc = (unsigned long )ret_from_kernel_thread; thread->cpu_context.r4 = stack_start; thread->cpu_context.r5 = stk_sz;
arch/x86/kernel/fpu/core.c +1 -1
@@ -91,7 +91,7 @@ void kernel_fpu_begin(void)this_cpu_write (in_kernel_fpu, true); -if (!(current->flags &PF_KTHREAD ) && +if (!is_kthread (current) && !test_thread_flag (TIF_NEED_FPU_LOAD )) {set_thread_flag (TIF_NEED_FPU_LOAD );/*
arch/x86/kernel/process_32.c +1 -1
@@ -134,7 +134,7 @@ int copy_thread_tls(unsigned long clone_flags, unsigned long sp, p->thread.sp0 = (unsigned long ) (childregs+1 );memset (p->thread.ptrace_bps,0 ,sizeof (p->thread.ptrace_bps)); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (childregs,0 ,sizeof (struct pt_regs )); frame->bx = sp;/* function */
arch/x86/kernel/process_64.c +1 -1
@@ -397,7 +397,7 @@ int copy_thread_tls(unsigned long clone_flags, unsigned long sp,savesegment (ds, p->thread.ds);memset (p->thread.ptrace_bps,0 ,sizeof (p->thread.ptrace_bps)); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {/* kernel thread */ memset (childregs,0 ,sizeof (struct pt_regs )); frame->bx = sp;/* function */
arch/xtensa/kernel/process.c +1 -1
@@ -217,7 +217,7 @@ int copy_thread(unsigned long clone_flags, unsigned long usp_thread_fn, p->thread.sp = (unsigned long )childregs; -if (!(p->flags &PF_KTHREAD )) { +if (!is_kthread (p)) {struct pt_regs *regs =current_pt_regs ();unsigned long usp = usp_thread_fn ? usp_thread_fn : regs->areg[1 ];
block/blk-cgroup.c +1 -1
@@ -1739,7 +1739,7 @@ void blkcg_maybe_throttle_current(void)*/ void blkcg_schedule_throttle (struct request_queue *q,bool use_memdelay) { -if (unlikely (current->flags &PF_KTHREAD )) +if (unlikely (is_kthread (current)))return ;if (!blk_get_queue (q))
drivers/tty/sysrq.c +1 -1
@@ -336,7 +336,7 @@ static void send_sig_all(int sig)read_lock (&tasklist_lock);for_each_process (p) { -if (p->flags &PF_KTHREAD ) +if (is_kthread (p))continue ;if (is_global_init (p))continue ;
fs/coredump.c +1 -1
@@ -412,7 +412,7 @@ static int zap_threads(struct task_struct *tsk, struct mm_struct *mm,for_each_process (g) {if (g == tsk->group_leader)continue ; -if (g->flags &PF_KTHREAD ) +if (is_kthread (g))continue ;for_each_thread (g, p) {
fs/file_table.c +2 -2
@@ -335,7 +335,7 @@ void fput_many(struct file *file, unsigned int refs)if (atomic_long_sub_and_test (refs, &file->f_count)) {struct task_struct *task = current; -if (likely (!in_interrupt () && !(task->flags &PF_KTHREAD ))) { +if (likely (!in_interrupt () && !is_kthread (task))) {init_task_work (&file->f_u.fu_rcuhead, ____fput);if (!task_work_add (task, &file->f_u.fu_rcuhead, true))return ;
@@ -368,7 +368,7 @@ void __fput_sync(struct file *file) {if (atomic_long_dec_and_test (&file->f_count)) {struct task_struct *task = current; -BUG_ON (!(task->flags &PF_KTHREAD )); +BUG_ON (!is_kthread (task));__fput (file); } }
fs/namespace.c +1 -1
@@ -1173,7 +1173,7 @@ static void mntput_no_expire(struct mount *mnt)if (likely (!(mnt->mnt.mnt_flags &MNT_INTERNAL ))) {struct task_struct *task = current; -if (likely (!(task->flags &PF_KTHREAD ))) { +if (likely (!is_kthread (task))) {init_task_work (&mnt->mnt_rcu, __cleanup_mnt);if (!task_work_add (task, &mnt->mnt_rcu, true))return ;
fs/proc/base.c +2 -2
@@ -1098,7 +1098,7 @@ static int __set_oom_adj(struct file *file, int oom_adj, bool legacy)continue ;/* do not touch kernel threads or the global init */ -if (p->flags &PF_KTHREAD ||is_global_init (p)) +if (is_kthread (p) ||is_global_init (p))continue ;task_lock (p);
@@ -1695,7 +1695,7 @@ void task_dump_owner(struct task_struct *task, umode_t mode,kuid_t uid;kgid_t gid; -if (unlikely (task->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (task))) { *ruid =GLOBAL_ROOT_UID ; *rgid =GLOBAL_ROOT_GID ;return ;
include/linux/cgroup.h +1 -1
@@ -907,7 +907,7 @@ static inline bool cgroup_task_freeze(struct task_struct *task) {bool ret; -if (task->flags &PF_KTHREAD ) +if (is_kthread (task))return false;rcu_read_lock ();
kernel/cgroup/freezer.c +2 -2
@@ -199,7 +199,7 @@ static void cgroup_do_freeze(struct cgroup *cgrp, bool freeze)* Ignore kernel threads here. Freezing cgroups containing * kthreads isn't supported. */ -if (task->flags &PF_KTHREAD ) +if (is_kthread (task))continue ;cgroup_freeze_task (task, freeze); }
@@ -227,7 +227,7 @@ void cgroup_freezer_migrate_task(struct task_struct *task,/* * Kernel threads are not supposed to be frozen at all. */ -if (task->flags &PF_KTHREAD ) +if (is_kthread (task))return ;/*
kernel/events/core.c +1 -1
@@ -5951,7 +5951,7 @@ static void perf_sample_regs_user(struct perf_regs *regs_user,if (user_mode (regs)) { regs_user->abi =perf_reg_abi (current); regs_user->regs = regs; - }else if (!(current->flags &PF_KTHREAD )) { + }else if (!is_kthread (current)) {perf_get_regs_user (regs_user, regs, regs_user_copy); }else { regs_user->abi =PERF_SAMPLE_REGS_ABI_NONE ;
kernel/exit.c +1 -1
@@ -448,7 +448,7 @@ void mm_update_next_owner(struct mm_struct *mm)* Search through everything else, we should not get here often. */ for_each_process (g) { -if (g->flags &PF_KTHREAD ) +if (is_kthread (g))continue ;for_each_thread (g, c) {if (c->mm == mm)
kernel/fork.c +3 -3
@@ -459,7 +459,7 @@ void free_task(struct task_struct *tsk)ftrace_graph_exit_task (tsk);put_seccomp_filter (tsk);arch_release_task_struct (tsk); -if (tsk->flags &PF_KTHREAD ) +if (is_kthread (tsk))free_kthread_struct (tsk);free_task_struct (tsk); }
@@ -1167,7 +1167,7 @@ struct file *get_task_exe_file(struct task_struct *task)task_lock (task); mm = task->mm;if (mm) { -if (!(task->flags &PF_KTHREAD )) +if (!is_kthread (task)) exe_file =get_mm_exe_file (mm); }task_unlock (task);
@@ -1191,7 +1191,7 @@ struct mm_struct *get_task_mm(struct task_struct *task)task_lock (task); mm = task->mm;if (mm) { -if (task->flags &PF_KTHREAD ) +if (is_kthread (task)) mm =NULL ;else mmget (mm);
kernel/freezer.c +2 -2
@@ -51,7 +51,7 @@ bool freezing_slow_path(struct task_struct *p)if (pm_nosig_freezing ||cgroup_freezing (p))return true; -if (pm_freezing && !(p->flags &PF_KTHREAD )) +if (pm_freezing && !is_kthread (p))return true;return false;
@@ -140,7 +140,7 @@ bool freeze_task(struct task_struct *p)return false; } -if (!(p->flags &PF_KTHREAD )) +if (!is_kthread (p))fake_signal_wake_up (p);else wake_up_state (p,TASK_INTERRUPTIBLE );
kernel/futex.c +1 -1
@@ -1249,7 +1249,7 @@ static int attach_to_pi_owner(u32 __user *uaddr, u32 uval, union futex_key *key,if (!p)return handle_exit_race (uaddr, uval,NULL ); -if (unlikely (p->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (p))) {put_task_struct (p);return -EPERM ; }
kernel/kthread.c +3 -3
@@ -72,7 +72,7 @@ static inline void set_kthread_struct(void *kthread)static inline struct kthread *to_kthread (struct task_struct *k) { -WARN_ON (!(k->flags &PF_KTHREAD )); +WARN_ON (!is_kthread (k));return (__force void *)k->set_child_tid; }
@@ -1205,7 +1205,7 @@ void kthread_associate_blkcg(struct cgroup_subsys_state *css) {struct kthread *kthread; -if (!(current->flags &PF_KTHREAD )) +if (!is_kthread (current))return ; kthread =to_kthread (current);if (!kthread)
@@ -1231,7 +1231,7 @@ struct cgroup_subsys_state *kthread_blkcg(void) {struct kthread *kthread; -if (current->flags &PF_KTHREAD ) { +if (is_kthread (current)) { kthread =to_kthread (current);if (kthread)return kthread->blkcg_css;
kernel/livepatch/transition.c +1 -1
@@ -358,7 +358,7 @@ static void klp_send_signals(void)* Meanwhile the task could migrate itself and the action * would be meaningless. It is not serious though. */ -if (task->flags &PF_KTHREAD ) { +if (is_kthread (task)) {/* * Wake up a kthread which sleeps interruptedly and * still has not been migrated.
kernel/ptrace.c +1 -1
@@ -376,7 +376,7 @@ static int ptrace_attach(struct task_struct *task, long request,audit_ptrace (task); retval = -EPERM ; -if (unlikely (task->flags &PF_KTHREAD )) +if (unlikely (is_kthread (task))) goto out;if (same_thread_group (task, current)) goto out;
kernel/sched/core.c +4 -4
@@ -1323,7 +1323,7 @@ void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags)static inline bool is_per_cpu_kthread (struct task_struct *p) { -if (!(p->flags &PF_KTHREAD )) +if (!is_kthread (p))return false;if (p->nr_cpus_allowed !=1 )
@@ -1518,7 +1518,7 @@ static int __set_cpus_allowed_ptr(struct task_struct *p, rq =task_rq_lock (p, &rf);update_rq_clock (rq); -if (p->flags &PF_KTHREAD ) { +if (is_kthread (p)) {/* * Kernel threads are allowed on online && !active CPUs */
@@ -1544,7 +1544,7 @@ static int __set_cpus_allowed_ptr(struct task_struct *p,do_set_cpus_allowed (p, new_mask); -if (p->flags &PF_KTHREAD ) { +if (is_kthread (p)) {/* * For kernel threads that do indeed end up on online && * !active we want to ensure they are strict per-CPU threads.
@@ -6649,7 +6649,7 @@ void normalize_rt_tasks(void)/* * Only normalize user tasks: */ -if (p->flags &PF_KTHREAD ) +if (is_kthread (p))continue ; p->se.exec_start =0 ;
kernel/sched/idle.c +1 -1
@@ -321,7 +321,7 @@ void play_idle(unsigned long duration_ms)*/ WARN_ON_ONCE (current->policy !=SCHED_FIFO );WARN_ON_ONCE (current->nr_cpus_allowed !=1 ); -WARN_ON_ONCE (!(current->flags &PF_KTHREAD )); +WARN_ON_ONCE (!is_kthread (current));WARN_ON_ONCE (!(current->flags &PF_NO_SETAFFINITY ));WARN_ON_ONCE (!duration_ms);
kernel/sched/wait.c +1 -1
@@ -380,7 +380,7 @@ EXPORT_SYMBOL(autoremove_wake_function);static inline bool is_kthread_should_stop (void ) { -return (current->flags &PF_KTHREAD ) &&kthread_should_stop (); +return is_kthread (current) &&kthread_should_stop (); }/*
kernel/signal.c +1 -1
@@ -1085,7 +1085,7 @@ static int __send_signal(int sig, struct kernel_siginfo *info, struct task_struc/* * Skip useless siginfo allocation for SIGKILL and kernel threads. */ -if ((sig ==SIGKILL ) || (t->flags &PF_KTHREAD )) +if ((sig ==SIGKILL ) ||is_kthread (t)) goto out_set;/*
kernel/stacktrace.c +1 -1
@@ -229,7 +229,7 @@ unsigned int stack_trace_save_user(unsigned long *store, unsigned int size)mm_segment_t fs;/* Trace user stack if not a kernel thread */ -if (current->flags &PF_KTHREAD ) +if (is_kthread (current))return 0 ; fs =get_fs ();
lib/is_single_threaded.c +1 -1
@@ -28,7 +28,7 @@ bool current_is_single_threaded(void) ret = false;rcu_read_lock ();for_each_process (p) { -if (unlikely (p->flags &PF_KTHREAD )) +if (unlikely (is_kthread (p)))continue ;if (unlikely (p == task->group_leader))continue ;
mm/memcontrol.c +1 -1
@@ -2668,7 +2668,7 @@ static void memcg_schedule_kmem_cache_create(struct mem_cgroup *memcg,static inline bool memcg_kmem_bypass (void ) { -if (in_interrupt () || !current->mm || (current->flags &PF_KTHREAD )) +if (in_interrupt () || !current->mm ||is_kthread (current))return true;return false; }
mm/oom_kill.c +2 -2
@@ -162,7 +162,7 @@ static bool oom_unkillable_task(struct task_struct *p) {if (is_global_init (p))return true; -if (p->flags &PF_KTHREAD ) +if (is_kthread (p))return true;return false; }
@@ -919,7 +919,7 @@ static void __oom_kill_process(struct task_struct *victim, const char *message)* No use_mm() user needs to read from the userspace so we are * ok to reap it. */ -if (unlikely (p->flags &PF_KTHREAD )) +if (unlikely (is_kthread (p)))continue ;do_send_sig_info (SIGKILL ,SEND_SIG_PRIV , p,PIDTYPE_TGID ); }
mm/page_alloc.c +1 -1
@@ -831,7 +831,7 @@ static inline struct capture_control *task_capc(struct zone *zone)struct capture_control *capc = current->capture_control;return capc && - !(current->flags &PF_KTHREAD ) && + !is_kthread (current) && !capc->page && capc->cc->zone == zone && capc->cc->direct_compaction ? capc :NULL ;
mm/vmacache.c +1 -1
@@ -30,7 +30,7 @@*/ static inline bool vmacache_valid_mm (struct mm_struct *mm) { -return current->mm == mm && !(current->flags &PF_KTHREAD ); +return current->mm == mm && !is_kthread (current); }void vmacache_update (unsigned long addr,struct vm_area_struct *newvma)
mm/vmscan.c +1 -1
@@ -3110,7 +3110,7 @@ static bool throttle_direct_reclaim(gfp_t gfp_mask, struct zonelist *zonelist,* committing a transaction where throttling it could forcing other * processes to block on log_wait_commit(). */ -if (current->flags &PF_KTHREAD ) +if (is_kthread (current)) goto out;/*
security/smack/smack_access.c +1 -1
@@ -671,7 +671,7 @@ bool smack_privileged(int cap)/* * All kernel tasks are privileged */ -if (unlikely (current->flags &PF_KTHREAD )) +if (unlikely (is_kthread (current)))return true;return smack_privileged_cred (cap,current_cred ());
security/smack/smack_lsm.c +2 -2
@@ -2257,7 +2257,7 @@ static int smack_sk_alloc_security(struct sock *sk, int family, gfp_t gfp_flags)/* * Sockets created by kernel threads receive web label. */ -if (unlikely (current->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (current))) { ssp->smk_in = &smack_known_web; ssp->smk_out = &smack_known_web; }else {
@@ -2761,7 +2761,7 @@ static int smack_socket_post_create(struct socket *sock, int family,/* * Sockets created by kernel threads receive web label. */ -if (unlikely (current->flags &PF_KTHREAD )) { +if (unlikely (is_kthread (current))) { ssp = sock->sk->sk_security; ssp->smk_in = &smack_known_web; ssp->smk_out = &smack_known_web;
security/yama/yama_lsm.c +1 -1
@@ -79,7 +79,7 @@ static void report_access(const char *access, struct task_struct *target,assert_spin_locked (&target->alloc_lock);/* for target->comm */ -if (current->flags &PF_KTHREAD ) { +if (is_kthread (current)) {/* I don't think kthreads call task_work_run() before exiting. * Imagine angry ranting about procfs here. */
Signature · 1 line
2.11.0
Re: [PATCH 2/9] sched: treewide: use is_kthread()
Hi Mark,
Quoted 24 lines · Mark Rutland
Now that we have
is_kthread(), let's convert existing open-coded checks of the form:
task->flags&PF_KTHREAD... over to the new helper, which makes things a little easier to read, and sets a consistent example for new code to follow.
Generated with coccinelle:
----
virtual patch@ depends on patch @
expression E;
@@- (
E->flags&PF_KTHREAD)
+is_kthread(E)
----... though this didn't pick up the instance in <
linux/cgroup.h>, which I fixed up manually.
Quoted 7 lines
Instances checking multiple PF_* flags at ocne are left as-is for now.
There should be no functional change as a result of this patch.
Signed-off-by Mark Rutland <[email protected]>arch/m68k/kernel/process.c | 2 +-
Gr{oetje,eeting}s,
Geert
Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected] In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds
Re: [PATCH 2/9] sched: treewide: use is_kthread()
In reply to Mark RutlandOn 2019-08-14 11:41:24 [+0100], Mark Rutland wrote:
…
Quoted 1 line
Instances checking multiple PF_* flags at ocne are left as-is for now.
s@ocne@once@
Sebastian
Mark Rutland
maintainerRe: [PATCH 2/9] sched: treewide: use is_kthread()
Quoted 7 lines · Sebastian Andrzej Siewior
On 2019-08-14 11:41:24 [+0100], Mark Rutland wrote:
…Quoted 1 line
Instances checking multiple PF_* flags at ocne are left as-is for now.
s@ocne@once@
Whoops; I'll fix that typo now...
Thanks,
Mark
Re: [PATCH 2/9] sched: treewide: use is_kthread()
In reply to Mark RutlandQuoted 8 lines · Mark Rutland
Now that we have
is_kthread(), let's convert existing open-coded checks of the form:
task->flags&PF_KTHREAD... over to the new helper, which makes things a little easier to read, and sets a consistent example for new code to follow.
[...]
Quoted 3 lines
kernel/sched/core.c | 8 ++++---- kernel/sched/idle.c | 2 +- kernel/sched/wait.c | 2 +-
Looks all sane there for me.
Mark Rutland
maintainerPATCH 3/9 kmemleak: correctly check for kthreads
In reply to Mark RutlandIn general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- mm/kmemleak.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
mm/kmemleak.c +3 -3
@@ -1290,10 +1290,10 @@ static int scan_should_stop(void)* This function may be called from either process or kthread context, * hence the need to check for both stop conditions. */ -if (current->mm) -return signal_pending (current); -else +if (is_kthread (current))return kthread_should_stop (); +else +return signal_pending (current);return 0 ; }
Signature · 1 line
2.11.0
Re: [PATCH 3/9] kmemleak: correctly check for kthreads
Quoted 7 lines · Mark Rutland
In general, a non-NULL
current->mmdoesn't imply that current is a kthread, as kthreads can install an mm viause_mm(), and so it's preferable to useis_kthread()to determine whether a thread is a kthread.For consistency, let's use
is_kthread()here.
Quoted 7 lines
Signed-off-by Mark Rutland <[email protected]>Cc Catalin Marinas <[email protected]>Cc Ingo Molnar <[email protected]>Cc Peter Zijlstra <[email protected]>--- mm/kmemleak.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
Quoted 4 lines
index
6e9e8cca663e..42ea3c14b577100644
mm/kmemleak.c static int scan_should_stop(void) · line 1296
Quoted 10 lines
mm/kmemleak.c +3
-3@@ -1290,10 +1290,10 @@ static int scan_should_stop(void)* This function may be called from either process or kthread context, * hence the need to check for both stop conditions. */ -if (current->mm) -return signal_pending (current); -else +if (is_kthread (current))return kthread_should_stop (); +else +return signal_pending (current);
While you are at it, you can drop the 'else' as well (occasionally checkpatch complains about 'else' after return).
Thanks.
Signature · 1 line
Catalin
Mark Rutland
maintainerPATCH 4/9 arm/perf: correctly check for kthreads
In reply to Mark RutlandThe arm perf_callchain_user() functions looks at current->mm, apparently to determine whether the thread is a kthread without any valid user context.
In general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- arch/arm/kernel/perf_callchain.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/kernel/perf_callchain.c +1 -1
@@ -71,7 +71,7 @@ perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regsperf_callchain_store (entry, regs->ARM_pc); -if (!current->mm) +if (is_kthread (current))return ; tail = (struct frame_tail __user *)regs->ARM_fp -1 ;
Signature · 1 line
2.11.0
Mark Rutland
maintainerPATCH 5/9 arm64: correctly check for ktrheads
In reply to Mark RutlandSince commit:
6eb6c80187c55b7f ("arm64: kernel thread don't need to save fpsimd context.")
... we skip saving the fpsimd state for kernel threads in arch_dup_task_struct(). We determine whether current is a kthread by looking at current->mm.
In general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- arch/arm64/kernel/process.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm64/kernel/process.c +1 -1
@@ -335,7 +335,7 @@ void arch_release_task_struct(struct task_struct *tsk)*/ int arch_dup_task_struct (struct task_struct *dst,struct task_struct *src) { -if (current->mm) +if (!is_kthread (current))fpsimd_preserve_current_state (); *dst = *src;
Signature · 1 line
2.11.0
Re: [PATCH 5/9] arm64: correctly check for ktrheads
Quoted 13 lines · Mark Rutland
Since commit:
6eb6c80187c55b7f("arm64: kernel thread don't need to save fpsimd context.")... we skip saving the fpsimd state for kernel threads in
arch_dup_task_struct(). We determine whether current is a kthread by looking atcurrent->mm.In general, a non-NULL
current->mmdoesn't imply that current is a kthread, as kthreads can install an mm viause_mm(), and so it's preferable to useis_kthread()to determine whether a thread is a kthread.
Quoted 7 lines
For consistency, let's use
is_kthread()here.Signed-off-by Mark Rutland <[email protected]>Cc Catalin Marinas <[email protected]>Cc Ingo Molnar <[email protected]>Cc Peter Zijlstra <[email protected]>Cc Will Deacon <[email protected]>
Mark Rutland
maintainerPATCH 6/9 sparc/perf: correctly check for kthreads
In reply to Mark RutlandThe sparc perf_callchain_user() functions looks at current->mm, apparently to determine whether the thread is a kthread without any valid user context.
In general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- arch/sparc/kernel/perf_event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/sparc/kernel/perf_event.c +1 -1
@@ -1858,7 +1858,7 @@ perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regsperf_callchain_store (entry, regs->tpc); -if (!current->mm) +if (is_kthread (current))return ;flushw_user ();
Signature · 1 line
2.11.0
Re: [PATCH 6/9] sparc/perf: correctly check for kthreads
From: Mark Rutland <[email protected]>
Date: Wed, 14 Aug 2019 11:41:28 +0100
Quoted 4 lines
The sparc
perf_callchain_user()functions looks atcurrent->mm, apparently to determine whether the thread is a kthread without any valid user context.
Quoted 8 lines
In general, a non-NULL
current->mmdoesn't imply that current is a kthread, as kthreads can install an mm viause_mm(), and so it's preferable to useis_kthread()to determine whether a thread is a kthread.For consistency, let's use
is_kthread()here.Signed-off-by Mark Rutland <[email protected]>
Mark Rutland
maintainerPATCH 7/9 x86/lbr: correctly check for kthreads
In reply to Mark RutlandThe x86 lbr branch_type() functions looks at current->mm, apparently to determine whether the thread is a kthread without any valid user context.
In general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- arch/x86/events/intel/lbr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/events/intel/lbr.c +1 -1
@@ -857,7 +857,7 @@ static int branch_type(unsigned long from, unsigned long to, int abort)* can happen if measuring at the user level only * and we interrupt in a kernel thread, e.g., idle. */ -if (!current->mm) +if (is_kthread (current))return X86_BR_NONE ;/* may fail if text not present */
Signature · 1 line
2.11.0
Mark Rutland
maintainerPATCH 8/9 x86/fpu: correctly check for kthreads
In reply to Mark RutlandPer commit:
0cecca9d03c964ab ("x86/fpu: Eager switch PKRU state")
... switch_fpu_state() is trying to distinguish user threads from kthreads, such that kthreads consistently use init_pkru_value. It does do by looking at current->mm.
In general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- arch/x86/include/asm/fpu/internal.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/include/asm/fpu/internal.h +1 -1
@@ -607,7 +607,7 @@ static inline void switch_fpu_finish(struct fpu *new_fpu)* PKRU state is switched eagerly because it needs to be valid before we * return to userland e.g. for a copy_to_user() operation. */ -if (current->mm) { +if (!is_kthread (current)) { pk =get_xsave_addr (&new_fpu->state.xsave,XFEATURE_PKRU );if (pk) pkru_val = pk->pkru;
Signature · 1 line
2.11.0
Re: [PATCH 8/9] x86/fpu: correctly check for kthreads
Quoted 4 lines · Mark Rutland
Per commit:
0cecca9d03c964ab("x86/fpu: Eager switch PKRU state")
Quoted 8 lines
...
switch_fpu_state()is trying to distinguish user threads from kthreads, such that kthreads consistently useinit_pkru_value. It does do by looking atcurrent->mm.In general, a non-NULL
current->mmdoesn't imply that current is a kthread, as kthreads can install an mm viause_mm(), and so it's preferable to useis_kthread()to determine whether a thread is a kthread.
I think this was missed in commit.
8d3289f2fa1e0 ("x86/fpu: Don't use current->mm to check for a kthread")A kthread with use_mm() would load here non-existing FPU state.
Sebastian
Mark Rutland
maintainerRe: [PATCH 8/9] x86/fpu: correctly check for kthreads
Quoted 9 lines · Sebastian Andrzej Siewior
Quoted 8 lines · Mark Rutland
Per commit:
0cecca9d03c964ab("x86/fpu: Eager switch PKRU state")...
switch_fpu_state()is trying to distinguish user threads from kthreads, such that kthreads consistently useinit_pkru_value. It does do by looking atcurrent->mm.
Quoted 7 lines
Quoted 4 lines
In general, a non-NULL
current->mmdoesn't imply that current is a kthread, as kthreads can install an mm viause_mm(), and so it's preferable to useis_kthread()to determine whether a thread is a kthread.I think this was missed in commit.
8d3289f2fa1e0 ("x86/fpu: Don't use current->mm to check for a kthread")
Yup, though if it's a bug it's been a bug since commit:
0cecca9d03c964ab ("x86/fpu: Eager switch PKRU state")
... which I guess the fixes tag would have to mention.
Quoted 4 lines
A kthread with
use_mm()would load here non-existing FPU state.
Given the above, should I add the fixes tag (for 0cecca9d03c964ab)?
Thanks,
Mark.
Re: [PATCH 8/9] x86/fpu: correctly check for kthreads
Quoted 7 lines · Mark Rutland
I think this was missed in commit.
Quoted 1 line
8d3289f2fa1e0 ("x86/fpu: Don't use current->mm to check for a kthread")Yup, though if it's a bug it's been a bug since commit:
0cecca9d03c964ab("x86/fpu: Eager switch PKRU state")
Quoted 8 lines
... which I guess the fixes tag would have to mention.
Quoted 4 lines
A kthread with
use_mm()would load here non-existing FPU state.Given the above, should I add the fixes tag (for
0cecca9d03c964ab)?
Buh. Either that or hch's commit (8d3289f2fa1e0 ("x86/fpu: Don't use current->mm to check for a kthread").
That will trigger a stable backport (for 5.2) asking what to do about is_kthread() so please leave a hint to use the PF_ flag like hch did.
Now that I think about it, even if we end up in kthread with use_mm() then its FPU state is non-existent which means get_xsave_addr() returns NULL. This is okay / expected but it triggers the WARN_ONCE().
Quoted 2 lines
Thanks,
Mark.
Sebastian
Mark Rutland
maintainerPATCH 9/9 samples/kretprobe: correctly check for kthreads
In reply to Mark RutlandIn general, a non-NULL current->mm doesn't imply that current is a kthread, as kthreads can install an mm via use_mm(), and so it's preferable to use is_kthread() to determine whether a thread is a kthread.
For consistency, let's use is_kthread() here.
--- samples/kprobes/kretprobe_example.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
samples/kprobes/kretprobe_example.c +1 -1
@@ -41,7 +41,7 @@ static int entry_handler(struct kretprobe_instance *ri, struct pt_regs *regs) {struct my_data *data; -if (!current->mm) +if (is_kthread (current))return 1 ;/* Skip kernel threads */ data = (struct my_data *)ri->data;
Signature · 1 line
2.11.0
Quoted 7 lines · Mark Rutland
A reasonable amount of kernel code looks at
task_struct::mmto determine whether a thread is a kthread or a real user task. This isn't quite right, since kthreads can have a non-NULL mm when callinguse_mm().The correct way to check whether a task is a kthread is to check whether
PF_KTHREADis set intask_struct::flags, but doing so is a bit unwieldy.
Quoted 9 lines
To make this a bit nicer, this series adds a new
is_kthread(tsk)helper, converts existing code to make use of it, and fixes up a number of erroneous checks ofcurrent->mm. Hopefully this will push people to use the right approach in future.I'm sure there are other instances in the kernel tree where we don't check this correctly. In this series I'm just trying to fix the instances I'm reasonably confident are incorrect.
I've only found one extra location in kernel/sched/ that seems like it wants to use is_kthread(), included below. I can send it out separately after this gets merged.
-----8<-----
kernel/sched/core.c +3 -4
@@ -1964,11 +1964,10 @@ static int select_fallback_rq(int cpu, struct task_struct *p) out:if (state != cpuset) {/* -* Don't tell them about moving exiting tasks or -* kernel threads (both mm NULL), since they never -* leave kernel. +* Don't tell them about moving exiting tasks (NULL mm) or +* kernel threads since they never leave kernel. */ -if (p->mm &&printk_ratelimit ()) { +if ((!is_kthread (p) && p->mm) &&printk_ratelimit ()) {printk_deferred ("process %d (%s) no longer affine to cpu%d\n" ,task_pid_nr (p), p->comm, cpu); }
----->8-----
Regarding sched/fair.c, I am quite convinced all of the p->mm checks are genuine (as in they don't care if it's a kthread or not) - those are all about NUMA balancing. Maybe Rik/Mel can correct me if I'm wrong.
Cheers,
Valentin