先生/女士：
您好！
为解决广大客户财务票据不足与减少税收.我公司有全国
各地发票优惠向外代开:(国税/地税)、普通商业销售、服务、
广告、建筑、运输、咨询、税务代开、另有海关代征增值税等
绝对保真,可先验证后再付款。
电话: 135 3783 2825
联 系 人: 刘先生
请留号码以备急用,
-
This SF.net email i
The current remapping algortithm assumes that each PT_LOAD elf segment has a
starting address (vaddr) that is aligned to the huge page size. This
condition was enforced by the custom linker scripts that are packaged with
libhugetlbfs.
Unaligned starting addresses can be corrected simply by aligni
The current implementation of libhugetlbfs segment remapping requires that a
binary be linked with a custom linker script that is included in the library
distribution. Over time these custom linker scripts have needed updating to
keep them compatible with changes occuring in binutils. While we've
The segment parsing code for binaries linked with custom linker scripts is now
nearly identical to the new system script method so we can consolidate two
functions into one code path.
There are two interesting semantics to note:
1) verify_segment_layout() will be called when we are using custom li
With the core of libhugetlbfs now supporting relinking without the need for
custom scripts, export the feature to users in an easy to use manner. For a
binary to be properly linked to support this remapping method, three things
must be added to the ld command line during the final link:
-zcommon-p
With all of the dependencies on custom linker scripts removed, introduce a
segment remapping algorithm that is compatible with the system linker scripts
(binutils 2.18 or later).
The only thing that is strictly required for full segment remapping to be
possible is a large enough gap between the te
On Thu, 2008-03-06 at 10:03 -0800, Nishanth Aravamudan wrote:
> What do folks think about this? I'd like to see the test-suite run as
> root and non-root successfully, given appropriate permissions on the
> hugetlbfs mount in 1.3
This all seems sound. Have you considered the case where the test
On Wed, 2008-03-12 at 10:39 -0500, Jon Tollefson wrote:
> morecore: The offset into the mmap needs to be a long to handle larger
> page sizes. This allows the "HUGETLB_MORECORE=yes" tests to pass with
> 16G huge pages.
>
> In trying to understand the code found the variables names "i" and "j"
>
In the direct test, we use a temporary file (/tmp/direct) along with a
hugetlbfs file to initiate direct I/O. Since we don't want to step on another
program that may be using a file with the same name, we open /tmp/direct with
O_EXCL to ensure it doesn't already exist. Unfortunately, we were forg
From: None
It seems that glibc symbols do not always have GLIBC in their names. This is
definitely the case on my Ubuntu Gutsy Gibbon box. Therefore, this mechanism
cannot be used to filter out non-glibc symbols. My guess is we need to use the
.gnu.version dependency-tracking if we want to rel
10 matches
Mail list logo
