linux-kernel - Re: [PATCH] binfmt_elf_fdpic: fix /proc/<pid>/auxv

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 for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAMo8Bf+GWXxHVorNvE=UWQDXfRvLE1MsK-dRwh_aZrd=ARxm2w@mail.gmail.com>
Date: Tue, 13 Aug 2024 11:28:06 -0700
From: Max Filippov <jcmvbkbc@...il.com>
To: Greg Ungerer <gregungerer@...tnet.com.au>
Cc: linux-kernel@...r.kernel.org, linux-mm@...ck.org, 
	linux-fsdevel@...r.kernel.org, Eric Biederman <ebiederm@...ssion.com>, 
	Kees Cook <keescook@...omium.org>, Alexander Viro <viro@...iv.linux.org.uk>, 
	Christian Brauner <brauner@...nel.org>, Jan Kara <jack@...e.cz>
Subject: Re: [PATCH] binfmt_elf_fdpic: fix /proc/<pid>/auxv

On Mon, Aug 12, 2024 at 9:53 PM Greg Ungerer <gregungerer@...tnet.com.au> wrote:
> On 13/8/24 04:02, Max Filippov wrote:
> > On Sun, Aug 11, 2024 at 7:26 PM Greg Ungerer <gregungerer@...tnet.com.au> wrote:
> >> On 23/3/24 05:54, Max Filippov wrote:
> >>> Althought FDPIC linux kernel provides /proc/<pid>/auxv files they are
> >>> empty because there's no code that initializes mm->saved_auxv in the
> >>> FDPIC ELF loader.
> >>>
> >>> Synchronize FDPIC ELF aux vector setup with ELF. Replace entry-by-entry
> >>> aux vector copying to userspace with initialization of mm->saved_auxv
> >>> first and then copying it to userspace as a whole.
> >>>
> >>> Signed-off-by: Max Filippov <jcmvbkbc@...il.com>
> >>
> >> This is breaking ARM nommu builds supporting fdpic and elf binaries for me.
> >>
> >> Tests I have for m68k and riscv nommu setups running elf binaries
> >> don't show any problems - I am only seeing this on ARM.
> >>
> >>
> >> ...
> >> Freeing unused kernel image (initmem) memory: 472K
> >> This architecture does not have kernel memory protection.
> >> Run /init as init process
> >> Internal error: Oops - undefined instruction: 0 [#1] ARM
> >> Modules linked in:
> >> CPU: 0 PID: 1 Comm: init Not tainted 6.10.0 #1
> >> Hardware name: ARM-Versatile (Device Tree Support)
> >> PC is at load_elf_fdpic_binary+0xb34/0xb80
> >> LR is at 0x0
> >> pc : [<00109ce8>]    lr : [<00000000>]    psr: 80000153
> >> sp : 00823e40  ip : 00000000  fp : 00b8fee4
> >> r10: 009c9b80  r9 : 00b8ff80  r8 : 009ee800
> >> r7 : 00000000  r6 : 009f7e80  r5 : 00b8fedc  r4 : 00b87000
> >> r3 : 00b8fed8  r2 : 00b8fee0  r1 : 00b87128  r0 : 00b8fef0
> >> Flags: Nzcv  IRQs on  FIQs off  Mode SVC_32  ISA ARM  Segment none
> >> Control: 00091176  Table: 00000000  DAC: 00000000
> >> Register r0 information: non-slab/vmalloc memory
> >> Register r1 information: slab/vmalloc mm_struct start 00b87000 pointer offset 296 size 428
> >> Register r2 information: non-slab/vmalloc memory
> >> Register r3 information: non-slab/vmalloc memory
> >> Register r4 information: slab/vmalloc mm_struct start 00b87000 pointer offset 0 size 428
> >> Register r5 information: non-slab/vmalloc memory
> >> Register r6 information: slab/vmalloc kmalloc-32 start 009f7e80 pointer offset 0 size 32
> >> Register r7 information: non-slab/vmalloc memory
> >> Register r8 information: slab/vmalloc kmalloc-512 start 009ee800 pointer offset 0 size 512
> >> Register r9 information: non-slab/vmalloc memory
> >> Register r10 information: slab/vmalloc kmalloc-128 start 009c9b80 pointer offset 0 size 128
> >> Register r11 information: non-slab/vmalloc memory
> >> Register r12 information: non-slab/vmalloc memory
> >> Process init (pid: 1, stack limit = 0x(ptrval))
> >> Stack: (0x00823e40 to 0x00824000)
> >> 3e40: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3e60: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3e80: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3ea0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3ec0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3ee0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3f00: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3f20: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3f40: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3f60: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3f80: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3fa0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3fc0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3fe0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> Call trace:
> >>    load_elf_fdpic_binary from bprm_execve+0x1b4/0x488
> >>    bprm_execve from kernel_execve+0x154/0x1e4
> >>    kernel_execve from kernel_init+0x4c/0x108
> >>    kernel_init from ret_from_fork+0x14/0x38
> >> Exception stack(0x00823fb0 to 0x00823ff8)
> >> 3fa0:                                     ???????? ???????? ???????? ????????
> >> 3fc0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
> >> 3fe0: ???????? ???????? ???????? ???????? ???????? ????????
> >> Code: bad PC value
> >> ---[ end trace 0000000000000000 ]---
> >> note: init[1] exited with irqs disabled
> >> Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b
> >> ---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b ]---
> >>
> >>
> >> The code around that PC is:
> >>
> >>     109cd0:       e2833ff1        add     r3, r3, #964    @ 0x3c4
> >>     109cd4:       e5933000        ldr     r3, [r3]
> >>     109cd8:       e5933328        ldr     r3, [r3, #808]  @ 0x328
> >>     109cdc:       e5933084        ldr     r3, [r3, #132]  @ 0x84
> >>     109ce0:       e5843034        str     r3, [r4, #52]   @ 0x34
> >>     109ce4:       eafffdbc        b       1093dc <load_elf_fdpic_binary+0x228>
> >>     109ce8:       e7f001f2        .word   0xe7f001f2
> >>     109cec:       eb09471d        bl      35b968 <__stack_chk_fail>
> >>     109cf0:       e59f0038        ldr     r0, [pc, #56]   @ 109d30 <load_elf_fdpic_binary+0xb7c>
> >>     109cf4:       eb092f03        bl      355908 <_printk>
> >>     109cf8:       eafffdb7        b       1093dc <load_elf_fdpic_binary+0x228>
> >>
> >>
> >> Reverting just this change gets it working again.
> >>
> >> Any idea what might be going on?
> >
> > Other than that the layout of the aux vector has slightly changed I don't
> > see anything. I can take a look at what's going on if you can share the
> > reproducer.
>
> I build the test binary using a simple script:
>
>    https://github.com/gregungerer/simple-linux/blob/master/build-armnommu-linux-uclibc-fdpic.sh
>
> Run the resulting vmlinux and devicetree under qemu as per comments at top of that script.

Cool scripts, thanks. I was able to reproduce the issue, it's the
BUG_ON(csp != sp);
from the create_elf_fdpic_tables() that causes it. I'm still trying
to figure out how that happens.

-- 
Thanks.
-- Max