linux-kernel - Re: CONFIG_DEBUG_VIRTUAL breaks boot on x86-32

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]

Date:   Wed, 27 Mar 2019 04:26:07 -0600
From:   William Kucharski <william.kucharski@...cle.com>
To:     Meelis Roos <mroos@...ux.ee>
Cc:     Ralph Campbell <rcampbell@...dia.com>,
        LKML <linux-kernel@...r.kernel.org>, linux-mm@...ck.org
Subject: Re: CONFIG_DEBUG_VIRTUAL breaks boot on x86-32


The dmesg output you posted confirms that max_low_pfn is indeed 0x373fe, and it appears
that the value of phys_mem being checked mat be 0x3f401ff1, which translates to pfn 0x3f401,
at least if what's still in registers can be believed.

Since that is indeed greater than max_low_pfn, VIRTUAL_BUG_ON triggers:

    VIRTUAL_BUG_ON((phys_addr >> PAGE_SHIFT) > max_low_pfn);

Looking at the call stack of

    copy_strings_0x220
        __check_object_size+0xef

that looks to translate to this sequence:

copy_from_user(kaddr+offset, str, bytes_to_copy)
    check_copy_size(kaddr+offset, bytes_to_copy, FALSE)
        check_object_size(kaddr+offset, bytes_to_copy, FALSE)
            __check_object_size(kaddr+offset, bytes_to_copy, FALSE)
                check_heap_object(kaddr+offset, bytes_to_copy, FALSE)
                    virt_to_head_page(kaddr+offset)
                        virt_to_page(kaddr+offset)
                            pfn_to_page(kaddr+offset)
                                __pa(kaddr+offset)
                                    __phys_addr(kaddr+offset)

so it appears the address is "too high" for low memory at that point.