linux-kernel - Re: [PATCH 0/1] (Was: Linux 3.11-rc4)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20130809171200.GA8901@redhat.com>
Date:	Fri, 9 Aug 2013 19:12:00 +0200
From:	Oleg Nesterov <oleg@...hat.com>
To:	Frederic Weisbecker <fweisbec@...il.com>
Cc:	Linus Torvalds <torvalds@...ux-foundation.org>,
	Grazvydas Ignotas <notasas@...il.com>,
	Felipe Contreras <felipe.contreras@...il.com>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Ingo Molnar <mingo@...e.hu>,
	Denys Vlasenko <dvlasenk@...hat.com>
Subject: Re: [PATCH 0/1] (Was: Linux 3.11-rc4)

On 08/09, Frederic Weisbecker wrote:
>
> On Thu, Aug 08, 2013 at 08:15:21PM +0200, Oleg Nesterov wrote:
> > But probably we should move "attr.bp_len == HW_BREAKPOINT_LEN_1" check
> > from arch_build_bp_info() to its caller, arch_validate_hwbkpt_settings().
> >
> > Because:
> >
> > > But this bp_len
> > > should rather be used for range breakpoints on archs that support it.
> >
> > Yes, exactly, and we already have the patches for amd, so bp->len can
> > be actually != 1 but currently we can't support because it is checked
> > in arch_build_bp_info().
>
> Hmm, but how moving that to arch_validate_hwbkpt_seetings() would solve
> the issue?

Of course, this itself won't solve the issue, sorry for confusion.

I meant that arch_build_bp_info(X86_BREAKPOINT_EXECUTE) should not fail
if ->bp_len is wrong, just because (unless we add more complications) it
can't know if it is correct or not (if the hardware supports the range
EXECUTE bps).

arch_validate_hwbkpt_settings() does the additional checks anyway, and
more importantly it checks cpu_has_bpext/mask. So I think it should also
have the additional check for X86_BREAKPOINT_EXECUTE case.

> > @@ -265,15 +262,11 @@ static int arch_build_bp_info(struct per
> >  		break;
> >  	case HW_BREAKPOINT_X:
> >  		info->type = X86_BREAKPOINT_EXECUTE;
> > -		/*
> > -		 * x86 inst breakpoints need to have a specific undefined len.
> > -		 * But we still need to check userspace is not trying to setup
> > -		 * an unsupported length, to get a range breakpoint for example.
> > -		 */
> > -		if (bp->attr.bp_len == sizeof(long)) {
> > -			info->len = X86_BREAKPOINT_LEN_X;
> > -			return 0;
> > -		}
> > +		/* until we change tools/perf */
> > +		if (bp->attr.bp_len == sizeof(long))
> > +			bp->attr.bp_len = HW_BREAKPOINT_LEN_1;
>
> Too bad we need to keep that compatibility around.

Yes, agreed... Do you see a better approach?

And just in case, it is not that I think that this hack is much better
than "ignore bp_len" as you suggested before. It's up to you.

> Do you think this could be
> a problem for AMD range breakpoints?

Yes, this doesn't look exactly right if ->bp_len == 8 actually tries to
denote a range. But this is the temporary hack, and at least currently
info->mask is only used if len > LEN_8, so I hope this should be fine.

> We can also fix the tools, then may be we'll be able to remove the kernel hack
> compatibility in a few years.

Or perhaps even earlier ;) And, perhaps after we change the tools we
can add pr_warn() for this case.

> Oh I need to check other archs as well.

Yes, and I'm afraid that tools/perf needs some arch-dependant define for
HW_BREAKPOINT_X's attr.bp_len. Or perhaps attr.bp_len == 0 could mean
"choose the right length" ?

Oleg.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/