Re: [PATCH v3 4/4] arc: Use generic dump_stack_print_cmdline() implementation

[Helge Deller]

On 10/10/22 07:18, Vineet Gupta wrote:
> On 8/8/22 06:09, Helge Deller wrote:
> > The process program name and command line is now shown in generic code
> > in dump_stack_print_info(), so drop the arc-specific implementation.
> >
> > Signed-off-by: Helge Deller <deller@xxxxxx>
>
> But that info printing was added back in 2018 by e36df28f532f882.
> I don't think arc is using show_regs_print_info -> dump_stack_print_info yet.
> Or is there a different code path now which calls here ?

Right.
See patches #1 and #2 of this series which added this
info to dump_stack_print_info().


> > ---
> >   arch/arc/kernel/troubleshoot.c | 24 ------------------------
> >   1 file changed, 24 deletions(-)
> >
> > diff --git a/arch/arc/kernel/troubleshoot.c b/arch/arc/kernel/troubleshoot.c
> > index 7654c2e42dc0..9807e590ee55 100644
> > --- a/arch/arc/kernel/troubleshoot.c
> > +++ b/arch/arc/kernel/troubleshoot.c
> > @@ -51,29 +51,6 @@ static void print_regs_callee(struct callee_regs *regs)
> >           regs->r24, regs->r25);
> >   }
> >
> > -static void print_task_path_n_nm(struct task_struct *tsk)
> > -{
> > -    char *path_nm = NULL;
> > -    struct mm_struct *mm;
> > -    struct file *exe_file;
> > -    char buf[ARC_PATH_MAX];
> > -
> > -    mm = get_task_mm(tsk);
> > -    if (!mm)
> > -        goto done;
> > -
> > -    exe_file = get_mm_exe_file(mm);
> > -    mmput(mm);
> > -
> > -    if (exe_file) {
> > -        path_nm = file_path(exe_file, buf, ARC_PATH_MAX-1);
> > -        fput(exe_file);
> > -    }
> > -
> > -done:
> > -    pr_info("Path: %s\n", !IS_ERR(path_nm) ? path_nm : "?");
> > -}
> > -
> >   static void show_faulting_vma(unsigned long address)
> >   {
> >       struct vm_area_struct *vma;
> > @@ -176,7 +153,6 @@ void show_regs(struct pt_regs *regs)
> >        */
> >       preempt_enable();
>
> Maybe we remove preempt* as well now (perhaps as a follow up patch)
> since that was added by f731a8e89f8c78 "ARC: show_regs: lockdep:
> re-enable preemption" where show_regs -> print_task_path_n_nm ->
> mmput was triggering lockdep splat which is supposedly removed.

The patch series was dropped from Andrew's queue, because the kernel
test robot showed some issues:
https://lore.kernel.org/lkml/Yu59QdVpPgnXUnQC@xsang-OptiPlex-9020/
Maybe adding preempt_enable() in my patches would fix that -
sadly I haven't had time to follow up on this yet ...

Helge

> > -    print_task_path_n_nm(tsk);
> >       show_regs_print_info(KERN_INFO);
> >
> >       show_ecr_verbose(regs);