linux/arch/alpha/kernel/entry.S, branch v6.8

alpha: lazy FPU switching

2023-02-25T04:14:22Z

On each context switch we save the FPU registers on stack of old process and restore FPU registers from the stack of new one. That allows us to avoid doing that each time we enter/leave the kernel mode; however, that can get suboptimal in some cases. For one thing, we don't need to bother saving anything for kernel threads. For another, if between entering and leaving the kernel a thread gives CPU up more than once, it will do useless work, saving the same values every time, only to discard the saved copy as soon as it returns from switch_to(). Alternative solution: * move the array we save into from switch_stack to thread_info * have a (thread-synchronous) flag set when we save them * have another flag set when they should be restored on return to userland. * do *NOT* save/restore them in do_switch_stack()/undo_switch_stack(). * restore on the exit to user mode if the restore flag had been set. Clear both flags. * on context switch, entry to fork/clone/vfork, before entry into do_signal() and on entry into straced syscall save the registers and set the 'saved' flag unless it had been already set. * on context switch set the 'restore' flag as well. * have copy_thread() set both flags for child, so the registers would be restored once the child returns to userland. * use the saved data in setup_sigcontext(); have restore_sigcontext() set both flags and copy from sigframe to save area. * teach ptrace to look for FPU registers in thread_info instead of switch_stack. * teach isolated accesses to FPU registers (rdfpcr, wrfpcr, etc.) to check the 'saved' flag (under preempt_disable()) and work with the save area if it's been set; if 'saved' flag is found upon write access, set 'restore' flag as well. Signed-off-by: Al Viro Signed-off-by: Matt Turner

alpha: replace NR_SYSCALLS by NR_syscalls

2023-02-14T17:37:17Z

Reference to other arch likes x86_64 or arm64 to do this replacement. To solve compile error when using NR_syscalls in kernel[1]. [1] https://lore.kernel.org/all/202203270449.WBYQF9X3-lkp@intel.com/ Signed-off-by: Yang Yang Signed-off-by: Matt Turner

alpha: ret_from_fork can go straight to ret_to_user

2022-10-30T03:31:16Z

We only hit ret_from_fork when the child is meant to return to userland (since 2012 or so). Signed-off-by: Al Viro

alpha: syscall exit cleanup

2022-10-30T03:31:15Z

$ret_success consists of two insn + branch to ret_from_syscall. The thing is, those insns are identical to the ones immediately preceding ret_from_syscall... Signed-off-by: Al Viro

alpha: fix handling of a3 on straced syscalls

2022-10-30T03:31:15Z

For successful syscall that happens to return a negative, we want a3 set to 0, no matter whether it's straced or not. As it is, for straced case we leave the value it used to have on syscall entry. Easily fixed, fortunately... Signed-off-by: Al Viro

alpha: fix syscall entry in !AUDUT_SYSCALL case

2022-10-30T03:31:15Z

We only want to take the slow path if SYSCALL_TRACE or SYSCALL_AUDIT is set; on !AUDIT_SYSCALL configs the current tree hits it whenever _any_ thread flag (including NEED_RESCHED, NOTIFY_SIGNAL, etc.) happens to be set. Fixes: a9302e843944 "alpha: Enable system-call auditing support" Signed-off-by: Al Viro

alpha: add support for TIF_NOTIFY_SIGNAL

2020-12-12T16:16:34Z

Wire up TIF_NOTIFY_SIGNAL handling for alpha. Cc: linux-alpha@vger.kernel.org Signed-off-by: Jens Axboe

alpha: unify the glue for sigreturn-like syscalls

2018-08-17T01:03:35Z

Signed-off-by: Al Viro

alpha: use alpha_ni_syscall only for syscall zero

2018-08-17T01:02:17Z

Once upon a time it used to have a C part that printed a warning about unimplemented OSF syscalls. That's what it's been doing all over the OSF syscall range, while the native Linux syscall range uses sys_ni_syscall(). With those warnings about unimplemented OSF syscalls gone (circa 2.4), alpha_ni_syscall() has shrunk to that little bit of asm and the only reason it hasn't been replaced with sys_ni_syscall() everywhere is that extra twist needed in case of syscall #0. Let's keep it only for syscall #0 and replace the rest with sys_ni_syscall. And use sys_ni_syscall for "number out range" in ptraced-call case, as we'd been doing for normal codepath since 2.1.86... Acked-by: Richard Henderson Signed-off-by: Al Viro

alpha: get rid of pointless insn in ret_from_kernel_thread

2018-03-28T12:58:27Z

It used to clear a3, so that signal handling on return to userland would've passed zero r0 to do_work_pending(), preventing the syscall restart logics from triggering. It had been pointless all along, since we only go there after successful do_execve(). Which does clear regs->r0 on alpha, preventing the syscall restart logics just fine, no extra help needed. Good thing, that, since back in 2012 do_work_pending() has lost the second argument, shifting the registers used to pass that thing from a3 to a2. Commit that had done that adjusted the entry.S code accordingly, but missed that one. As the result, we were left with useless insn in ret_from_kernel_thread and confusing comment to go with it. Get rid of both... Signed-off-by: Al Viro