Messages in this thread | | | Date | Sat, 3 Jul 2021 17:22:05 +0200 | From | Pavel Machek <> | Subject | Re: [PATCH 4.4 41/57] x86/fpu: Reset state for all signal restore failures |
| |
Hi!
> commit efa165504943f2128d50f63de0c02faf6dcceb0d upstream. > > If access_ok() or fpregs_soft_set() fails in __fpu__restore_sig() then the > function just returns but does not clear the FPU state as it does for all > other fatal failures. > > Clear the FPU state for these failures as well.
That's quite a confusing calling convention, right?
> +++ b/arch/x86/kernel/fpu/signal.c > @@ -262,15 +262,23 @@ static int __fpu__restore_sig(void __user *buf, void __user *buf_fx, int size) > return 0; > } > > - if (!access_ok(VERIFY_READ, buf, size)) > + if (!access_ok(VERIFY_READ, buf, size)) { > + fpu__clear(fpu); > return -EACCES; > + }
This returns -errno on failure.
> fpu__activate_curr(fpu); > > - if (!static_cpu_has(X86_FEATURE_FPU)) > - return fpregs_soft_set(current, NULL, > - 0, sizeof(struct user_i387_ia32_struct), > - NULL, buf) != 0; > + if (!static_cpu_has(X86_FEATURE_FPU)) { > + int ret = fpregs_soft_set(current, NULL, 0, > + sizeof(struct user_i387_ia32_struct), > + NULL, buf); > + > + if (ret) > + fpu__clear(fpu); > + > + return ret != 0; > + }
And this explicitely turns negative error into 1 for maximum confusion.
I don't see the same confusion in mainline.
Best regards, Pavel -- DENX Software Engineering GmbH, Managing Director: Wolfgang Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
[unhandled content-type:application/pgp-signature] | |