From mboxrd@z Thu Jan 1 00:00:00 1970 From: catalin.marinas@arm.com (Catalin Marinas) Date: Mon, 23 Mar 2015 12:02:54 +0000 Subject: [PATCH] tile: use si_int instead of si_ptr for compat_siginfo In-Reply-To: <201503161908.t2GJ8fs5021877@farm-0002.internal.tilera.com> References: <1423563011-12377-1-git-send-email-bamvor.zhangjian@huawei.com> <20150210122718.GC32052@e104818-lin.cambridge.arm.com> <54DB3B60.4050100@huawei.com> <20150211154054.GD9058@e104818-lin.cambridge.arm.com> <54DDAF2B.2070707@huawei.com> <20150213104455.GA3508@e104818-lin.cambridge.arm.com> <54DE730D.3090100@ezchip.com> <20150214112220.GB10246@MBP.local> <54ECF309.3020509@ezchip.com> <201503161908.t2GJ8fs5021877@farm-0002.internal.tilera.com> Message-ID: <20150323120253.GA12757@e104818-lin.cambridge.arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Mon, Mar 16, 2015 at 03:04:05PM -0400, Chris Metcalf wrote: > To be compatible with the generic get_compat_sigevent(), the > copy_siginfo_to_user32() and thus copy_siginfo_from_user32() > have to use si_int instead of si_ptr. Using si_ptr means that > for the case of ILP32 compat code running in big-endian mode, > we would end up copying the high 32 bits of the pointer value > into si_int instead of the desired low 32 bits. > > Signed-off-by: Chris Metcalf > Cc: Catalin Marinas > --- > arch/tile/kernel/compat_signal.c | 9 +++------ > 1 file changed, 3 insertions(+), 6 deletions(-) > > diff --git a/arch/tile/kernel/compat_signal.c b/arch/tile/kernel/compat_signal.c > index 8c5abf2e4794..bca13054afb4 100644 > --- a/arch/tile/kernel/compat_signal.c > +++ b/arch/tile/kernel/compat_signal.c > @@ -68,7 +68,7 @@ int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *fr > if (from->si_code < 0) { > err |= __put_user(from->si_pid, &to->si_pid); > err |= __put_user(from->si_uid, &to->si_uid); > - err |= __put_user(ptr_to_compat(from->si_ptr), &to->si_ptr); > + err |= __put_user(from->si_int, &to->si_int); > } else { > /* > * First 32bits of unions are always present: > @@ -93,8 +93,7 @@ int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *fr > break; > case __SI_TIMER >> 16: > err |= __put_user(from->si_overrun, &to->si_overrun); > - err |= __put_user(ptr_to_compat(from->si_ptr), > - &to->si_ptr); > + err |= __put_user(from->si_int, &to->si_int); > break; It's usually the __SI_TIMER and __SI_MESGQ cases that matters here (the latter already handled). I'm not entirely sure about the si_code < 0 change. > /* This is not generated by the kernel as of now. */ > case __SI_RT >> 16: > @@ -110,7 +109,6 @@ int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *fr > int copy_siginfo_from_user32(siginfo_t *to, struct compat_siginfo __user *from) > { > int err; > - u32 ptr32; > > if (!access_ok(VERIFY_READ, from, sizeof(struct compat_siginfo))) > return -EFAULT; > @@ -121,8 +119,7 @@ int copy_siginfo_from_user32(siginfo_t *to, struct compat_siginfo __user *from) > > err |= __get_user(to->si_pid, &from->si_pid); > err |= __get_user(to->si_uid, &from->si_uid); > - err |= __get_user(ptr32, &from->si_ptr); > - to->si_ptr = compat_ptr(ptr32); > + err |= __get_user(to->si_int, &from->si_int); We have a memset(to, 0, sizeof(*to)) on arm64 in this function but I can't see it on tile. Some members or even half of si_ptr would be left uninitialised. -- Catalin