Message-ID: <20170104142526.wudj4ub2yzmc3bng@redhat.com>
Date:   Wed, 4 Jan 2017 09:25:27 -0500
From:   Peter Jones <pjones@...hat.com>
To:     Dan Williams <dan.j.williams@...el.com>
Cc:     Matt Fleming <matt@...eblueprint.co.uk>,
        Ingo Molnar <mingo@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        "H . Peter Anvin" <hpa@...or.com>,
        Ard Biesheuvel <ard.biesheuvel@...aro.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-efi@...r.kernel.org, Borislav Petkov <bp@...en8.de>,
        Dave Young <dyoung@...hat.com>,
        Leif Lindholm <leif.lindholm@...aro.org>,
        Mark Rutland <mark.rutland@....com>,
        "Verma, Vishal L" <vishal.l.verma@...el.com>, poros@...hat.com
Subject: Re: [PATCH 08/29] efi: Allow drivers to reserve boot services forever

On Tue, Jan 03, 2017 at 06:48:43PM -0800, Dan Williams wrote:
> On Fri, Sep 9, 2016 at 8:18 AM, Matt Fleming <matt@...eblueprint.co.uk> wrote:
> > Today, it is not possible for drivers to reserve EFI boot services for
> > access after efi_free_boot_services() has been called on x86. For
> > ARM/arm64 it can be done simply by calling memblock_reserve().
> >
> > Having this ability for all three architectures is desirable for a
> > couple of reasons,
> >
> >   1) It saves drivers copying data out of those regions
> >   2) kexec reboot can now make use of things like ESRT
> >
> > Instead of using the standard memblock_reserve() which is insufficient
> > to reserve the region on x86 (see efi_reserve_boot_services()), a new
> > API is introduced in this patch; efi_mem_reserve().
> >
> > efi.memmap now always represents which EFI memory regions are
> > available. On x86 the EFI boot services regions that have not been
> > reserved via efi_mem_reserve() will be removed from efi.memmap during
> > efi_free_boot_services().
> >
> > This has implications for kexec, since it is not possible for a newly
> > kexec'd kernel to access the same boot services regions that the
> > initial boot kernel had access to unless they are reserved by every
> > kexec kernel in the chain.
> >
> > Tested-by: Dave Young <dyoung@...hat.com> [kexec/kdump]
> > Tested-by: Ard Biesheuvel <ard.biesheuvel@...aro.org> [arm]
> > Acked-by: Ard Biesheuvel <ard.biesheuvel@...aro.org>
> > Cc: Leif Lindholm <leif.lindholm@...aro.org>
> > Cc: Peter Jones <pjones@...hat.com>
> > Cc: Borislav Petkov <bp@...en8.de>
> > Cc: Mark Rutland <mark.rutland@....com>
> > Signed-off-by: Matt Fleming <matt@...eblueprint.co.uk>
> 
> This commit appears to cause a boot regression between v4.8 and v4.9.

Can you verify that the memory map looks reasonable?  This looks similar
(but not identical) to the traceback I was hitting on some ThinkPads
with ESRT last month, so there's some chance, if it is caused by bad
memory map entries, it may be fixed by 1cb209f63 on efi/next .

> 
> BUG: unable to handle kernel paging request at ffff8830281bf1c8
> IP: [<ffffffff81a21266>] __next_mem_range_rev+0x13a/0x1d6
> PGD 3193067 PUD 3196067 PTE 80000030281bf060
> Oops: 0000 1 SMP DEBUG_PAGEALLOC
> Modules linked in:
> CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.9.0+ #2
> task: ffffffff82011540 task.stack: ffffffff82000000
> RIP: 0010:[<ffffffff81a21266>] [<ffffffff81a21266>]
> __next_mem_range_rev+0x13a/0x1d6
> RSP: 0000:ffffffff82003dd8 EFLAGS: 00010202
> RAX: ffff8830281bf1e0 RBX: ffffffff82003e60 RCX: ffffffff82167490
> RDX: 0000000000000000 RSI: 00000000ffffffff RDI: 0000001840000000
> RBP: ffffffff82003e18 R08: ffffffff821674b0 R09: 000000000000008f
> R10: 000000000000008f R11: ffffffff82011cf0 R12: 0000000000000004
> R13: 0000003040000000 R14: 0000000000000000 R15: 0000000000000001
> FS: 0000000000000000(0000) GS:ffff8817e0800000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: ffff8830281bf1c8 CR3: 000000000200a000 CR4: 00000000007406f0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> PKRU: 00000000
> Stack:
> ffffea0000001700 ffffffff82003e50 0000000000000000
> 00000000000010000000003040000000 ffffffff82003e58 0000000000000180
> ffffffffffffffc0
> ffffffff82003e98 ffffffff81a21395 ffffffff82003e58 0000000000000000
> Call Trace:
> [<ffffffff81a21395>] memblock_find_in_range_node+0x93/0x13a
> [<ffffffff8221e3b1>] memblock_alloc_range_nid+0x1b/0x3e
> [<ffffffff8221e5b9>] __memblock_alloc_base+0x15/0x17
> [<ffffffff8221e5cd>] memblock_alloc_base+0x12/0x2e
> [<ffffffff8221e5f4>] memblock_alloc+0xb/0xd
> [<ffffffff82208e22>] efi_free_boot_services+0x46/0x180
> [<ffffffff821e818b>] start_kernel+0x4a1/0x4cc
> [<ffffffff821e7ad8>] ? set_init_arg+0x55/0x55
> [<ffffffff821e7120>] ? early_idt_handler_array+0x120/0x120
> [<ffffffff821e75d6>] x86_64_start_reservations+0x2a/0x2c
> [<ffffffff821e7724>] x86_64_start_kernel+0x14c/0x16f
> Code: 18 44 89 38 41 8d 44 24 ff 49 c1 e1 20 4c 09 c8 48 89 03 e9 a0
> 00 00 00 4d 63 d1 4c 89 d0 48 c1 e0 05 49 03 40 18 45 85 c9 74 28 <48>
> 8b 50 e8 48 03 50 e0 49 83 cb ff 4d 3b 10 73 03 4c 8b 18 49 ^M
> RIP [<ffffffff81a21266>] __next_mem_range_rev+0x13a/0x1d6
> 
> I also see that Petr may have run into it as well [1]? Petr is this
> the same signature you are seeing? Can you post a boot log with
> "efi=debug" on the kernel command line?
> 
> It also fails on 4.10-rc2.  However, if I revert the following commits
> it boots fine:
> 
> 4bc9f92e64c8 x86/efi-bgrt: Use efi_mem_reserve() to avoid copying image data
> 8e80632fb23f efi/esrt: Use efi_mem_reserve() and avoid a kmalloc()
> 816e76129ed5 efi: Allow drivers to reserve boot services forever
> 
> [1]: https://lkml.org/lkml/2016/12/21/197

-- 
        Peter

Powered by blists - more mailing lists