linux-kernel - [Patch v2 0/6] Refactor split_mem_range with proper helper and loop

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

Message-Id: <20191205021403.25606-1-richardw.yang@linux.intel.com>
Date:   Thu,  5 Dec 2019 10:13:57 +0800
From:   Wei Yang <richardw.yang@...ux.intel.com>
To:     x86@...nel.org, linux-kernel@...r.kernel.org
Cc:     richard.weiyang@...il.com, dave.hansen@...ux.intel.com,
        luto@...nel.org, peterz@...radead.or, tglx@...utronix.de,
        Wei Yang <richardw.yang@...ux.intel.com>
Subject: [Patch v2 0/6] Refactor split_mem_range with proper helper and loop

split_mem_range is used to prepare range before mapping kernel page table.

After first version, Thomas suggested some brilliant idea to re-write the
logic.

Wei split the big patch into pieces and did some tests.  To verify the
functionality, Wei abstract the code into userland and did following test
cases:
    
        * ranges fits only 4K
        * ranges fits only 2M
        * ranges fits only 1G
        * ranges fits 4K and 2M
        * ranges fits 2M and 1G
        * ranges fits 4K, 2M and 1G
        * ranges fits 4K, 2M and 1G but w/o 1G size
        * ranges fits 4K, 2M and 1G with only 4K size
    
    Below is the test result:
    
        ### Split [4K, 16K][0x00001000-0x00004000]:
        [mem 0x00001000-0x00003fff] page size 4K
        ### Split [4M, 64M][0x00400000-0x04000000]:
        [mem 0x00400000-0x03ffffff] page size 2M
        ### Split [0G, 2G][0000000000-0x80000000]:
        [mem 0000000000-0x7fffffff] page size 1G
        ### Split [16K, 4M + 16K][0x00004000-0x00404000]:
        [mem 0x00004000-0x001fffff] page size 4K
        [mem 0x00200000-0x003fffff] page size 2M
        [mem 0x00400000-0x00403fff] page size 4K
        ### Split [4M, 2G + 2M][0x00400000-0x80200000]:
        [mem 0x00400000-0x3fffffff] page size 2M
        [mem 0x40000000-0x7fffffff] page size 1G
        [mem 0x80000000-0x801fffff] page size 2M
        ### Split [4M - 16K, 2G + 2M + 16K][0x003fc000-0x80204000]:
        [mem 0x003fc000-0x003fffff] page size 4K
        [mem 0x00400000-0x3fffffff] page size 2M
        [mem 0x40000000-0x7fffffff] page size 1G
        [mem 0x80000000-0x801fffff] page size 2M
        [mem 0x80200000-0x80203fff] page size 4K
        ### Split w/o 1G size [4M - 16K, 2G + 2M + 16K][0x003fc000-0x80204000]:
        [mem 0x003fc000-0x003fffff] page size 4K
        [mem 0x00400000-0x801fffff] page size 2M
        [mem 0x80200000-0x80203fff] page size 4K
        ### Split w/ only 4K [4M - 16K, 2G + 2M + 16K][0x003fc000-0x80204000]:
        [mem 0x003fc000-0x80203fff] page size 4K

Thomas Gleixner (1):
  x86/mm: Refactor split_mem_range with proper helper and loop

Wei Yang (5):
  x86/mm: Remove second argument of split_mem_range()
  x86/mm: Add attribute __ro_after_init to after_bootmem
  x86/mm: Make page_size_mask unsigned int clearly
  x86/mm: Refine debug print string retrieval function
  x86/mm: Use address directly in split_mem_range()

 arch/x86/mm/init.c | 259 ++++++++++++++++++---------------------------
 1 file changed, 103 insertions(+), 156 deletions(-)

-- 
2.17.1