----- Original Message ----- > On Wed, Jun 22, 2016 at 01:35:02PM +0900, AKASHI Takahiro wrote: ... [ cut ] ... > > + if (!arm64_on_irq_stack(bt->tc->processor, fp)) > > Obviously I meant "if (arm64_on_irq_stack(bt->tc->processor,fp))" > > > + /* > > + * Very rare corner case: > > + * We have already exited an irq handler, > > + * and switched from IRQ stack to process > > + * stack but fp is not yet restored. > > + * Fp still points to IRQ stack, sigh. > > + * For example, > > + * ret_to_user() after irq_handler in el0_irq > > + */ > > + frame->sp = bt->stacktop > > + - sizeof(struct arm64_pt_regs) > > + - 0x10; > > Anyhow, this logic seems to still have some bug. > I will try to fix it. > > Sorry for any confusion. > -Takahiro AKASHI
No problem -- I'll await v5. Thanks, Dave -- Crash-utility mailing list [email protected] https://www.redhat.com/mailman/listinfo/crash-utility
