Archive formats such as .zip files are generally susceptible to
so-called "traversal attacks". This allows an attacker to craft
an archive that writes to unexpected locations of the file system
(e.g., /etc/shadow) if an unspecting root user were to unpack a
malicious archive.
This patch neutralizes absolute paths such as /tmp/moo and deeply
relative paths such as dummy/../../../../../../../../../../tmp/moo
The Debian project requested CVE-2014-9485 be allocated for the
first identified weakness. The fix was incomplete, resulting in a
revised patch applied here. Since there wasn't an updated version
released by Debian with the incomplete fix, I suggest we use this
CVE to identify both issues.
Link: https://security.snyk.io/research/zip-slip-vulnerability
Link: https://bugs.debian.org/774321
Link: https://bugs.debian.org/776831
Link: https://nvd.nist.gov/vuln/detail/CVE-2014-9485
Reported-by: Jakub Wilk <jwilk@debian.org>
Fixed-by: Michael Gilbert <mgilbert@debian.org>
The appnote says that if the number of entries in the end record
is 0xffff, then the actual number of entries will be found in the
Zip64 end record. Therefore if the number of entries is equal to
0xffff, it can't be in the end record by itself, since that is an
instruction to get the number from the Zip64 end record. This code
would just store 0xffff in the end record in that case, not making
a Zip64 end record. This commit fixes that.
A bug fix in zlib 1.2.12 resulted in a slight slowdown (1-2%) of
deflate. This commit provides the option to #define LIT_MEM, which
uses more memory to reverse most of that slowdown. The memory for
the pending buffer and symbol buffers is increased by 25%, which
increases the total memory usage with the default parameters by
about 6%.
This checks the lengths of the file name, extra field, and comment
that would be put in the zip headers, and rejects them if they are
too long. They are each limited to 65535 bytes in length by the zip
format. This also avoids possible buffer overflows if the provided
fields are too long.
slide_hash() knowingly reads potentially uninitialized memory, see
comment lower down about prev[n] potentially being garbage. In
this case, the result is never used.
This also adds --address for the address sanitizer, in addition to
the existing --sanitizer. -fno-omit-frame-pointer has been added
for both sanitizers to improve the error reporting.
