linux-kernel - Re: BUG: unable to handle kernel NULL pointer dereference in handle_external_interrupt

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

Date:   Mon, 23 Mar 2020 17:31:15 +0100
From:   Alexander Potapenko <glider@...gle.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     Dmitry Vyukov <dvyukov@...gle.com>,
        syzbot <syzbot+3f29ca2efb056a761e38@...kaller.appspotmail.com>,
        clang-built-linux <clang-built-linux@...glegroups.com>,
        Borislav Petkov <bp@...en8.de>,
        "H. Peter Anvin" <hpa@...or.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>, KVM list <kvm@...r.kernel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Ingo Molnar <mingo@...hat.com>,
        "Christopherson, Sean J" <sean.j.christopherson@...el.com>,
        syzkaller-bugs <syzkaller-bugs@...glegroups.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Vitaly Kuznetsov <vkuznets@...hat.com>, wanpengli@...cent.com,
        "the arch/x86 maintainers" <x86@...nel.org>
Subject: Re: BUG: unable to handle kernel NULL pointer dereference in handle_external_interrupt_irqoff

On Mon, Mar 23, 2020 at 9:18 AM Paolo Bonzini <pbonzini@...hat.com> wrote:
>
> On 22/03/20 07:59, Dmitry Vyukov wrote:
> >
> > The commit range is presumably
> > fb279f4e238617417b132a550f24c1e86d922558..63849c8f410717eb2e6662f3953ff674727303e7
> > But I don't see anything that says "it's me". The only commit that
> > does non-trivial changes to x86/vmx seems to be "KVM: VMX: check
> > descriptor table exits on instruction emulation":
>
> That seems unlikely, it's a completely different file and it would only
> affect the outside (non-nested) environment rather than your own kernel.
>
> The only instance of "0x86" in the registers is in the flags:
>
> > RSP: 0018:ffffc90001ac7998 EFLAGS: 00010086
> > RAX: ffffc90001ac79c8 RBX: fffffe0000000000 RCX: 0000000000040000
> > RDX: ffffc9000e20f000 RSI: 000000000000b452 RDI: 000000000000b453
> > RBP: 0000000000000ec0 R08: ffffffff83987523 R09: ffffffff811c7eca
> > R10: ffff8880a4e94200 R11: 0000000000000002 R12: dffffc0000000000
> > R13: fffffe0000000ec8 R14: ffffffff880016f0 R15: fffffe0000000ecb
> > FS:  00007fb50e370700(0000) GS:ffff8880ae800000(0000) knlGS:0000000000000000
> > CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > CR2: 000000000000005c CR3: 0000000092fc7000 CR4: 00000000001426f0
> > DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> > DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>
> That would suggest a miscompilation of the inline assembly, which does
> push the flags:
>
> #ifdef CONFIG_X86_64
>                 "mov %%" _ASM_SP ", %[sp]\n\t"
>                 "and $0xfffffffffffffff0, %%" _ASM_SP "\n\t"
>                 "push $%c[ss]\n\t"
>                 "push %[sp]\n\t"
> #endif
>                 "pushf\n\t"
>                 __ASM_SIZE(push) " $%c[cs]\n\t"
>                 CALL_NOSPEC
>
>
> It would not explain why it suddenly started to break, unless the clang
> version also changed, but it would be easy to ascertain and fix (in
> either KVM or clang).  Dmitry, can you send me the vmx.o and
> kvm-intel.ko files?

On a quick glance, Clang does not miscompile this part.
Attached is the disassembly of handle_external_interrupt_irqoff() from
v5.4 (where the problem seems to also reproduce) with Clang and GCC.
They do virtually the same (look for asm blob after kvm_before_interrupt()).

View attachment "handle_external_interrupt_irqoff.gcc.txt" of type "text/plain" (14916 bytes)

View attachment "handle_external_interrupt_irqoff.clang.txt" of type "text/plain" (15167 bytes)