aboutsummaryrefslogtreecommitdiff
path: root/lib
AgeCommit message (Collapse)Author
2019-09-12Cleanup: move block processor code to sub directoryDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-11Cleanup Automake file for librariesDavid Oberhollenzer
Split the signel file up into several small ones and use a variable for the public headers instead of duplicating them. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-10Minor cleanup of the block processorDavid Oberhollenzer
- Set the "compression failed" flag in the common implementation. - Use static initializers for the pthread code. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-10Make the thread pool queue backlog user configurableDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-10API cleanup: eliminate boolean argumentsDavid Oberhollenzer
Replace them with flag fields which can be extended in the future without breaking the ABI. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-10Cleanup: mark arguments const where possibleDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-09Change license of libsquashfs.soDavid Oberhollenzer
This this commit, the code that is compiled and linked to produce libsquashfs.so (and *only* libsquashfs.so) is released under the terms and conditions of the GNU Lesser General Public license, allowing 3rd party programs to dynamically link against it and still being able to release their programs under a license of their choosing. The libutil library is also relicensed, because it is statically linked into libsquashfs.so. This does not affect the command line programs that remain under the GPLv3. At the current time, the two libraries do not contain any 3rd party contributions and thus do not require anybodys consent other than mine to change the licensing terms (unlike e.g. libfstree.a). The libfstree.a and libsqfshelper.a utility libraries will also remain under the GPLv3 license. Going forward, some parts of libsqfshelper.a will be moved to libsquashfs.so. The parts absorbed by libsquashfs.so will then also have their licensing conditions changed *at that time*. For the public headers, this commit adds the full boiler-plate comment about the license, for the source files and internal headers, only the SPDX identifier is changed. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-09Remove ad-hoc image size accountingDavid Oberhollenzer
Much simpler to just set the correct size in the end, before writing the super block. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-09Move file name sanity check to deserialize_fstreeDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-08Replace direct file I/O with abstraction layerDavid Oberhollenzer
This should make it easier to use libsquashfs with custom setups that embedd a squashfs image inside something else. Also, it should make it easier to port to non unix-like platforms. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-07Remove printing to stderr in libsquashfs with returning error numbersDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-07Move data structure forward declarations to predef headerDavid Oberhollenzer
This in turn allows for removing most of the header inclusions within the public headers. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Fix "safe" string allocation wrapperDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Fix API visibility for libsquashfs.soDavid Oberhollenzer
This commit adds a "predef.h" header to libsquashfs that exposes two macros, SQFS_API and SQFS_INTERNAL that are used to explicilty set the visibiliy of stuff that ends up in libsquashfs to default or hidden. For compatibillity with Windows which may at some point eventually be considered, the SQFS_API define is set to dllexport IF the cpp symbol SQFS_BUILDING_DLL is defined and set to import otherwise. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup naming scheme of compressorDavid Oberhollenzer
Add sqfs_* prefix to compressor, move implementation prefix up front. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup: add SQFS_*/sqfs_* prefix to block processorDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup: add sqfs_* prefix to id tableDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup: add sqfs_* prefix to meta readerDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup: add sqfs_* prefix to meta writerDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup: unify sqfs/SQFS name prefix for xattr readerDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Move block processor from compressor to libsquashfs codeDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Minor cleanup: remove if (...) goto skip construct from data readerDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-05Cleanup: move "create_block" from block processor over to data writerDavid Oberhollenzer
It only has one user and is quite specialized actually. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-04Store number of file blocks in generic inode structDavid Oberhollenzer
This simplifies things a little bit. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-04Split fstree inode serialization, move independend part to libsquashfs.soDavid Oberhollenzer
This commit adds a function to libsquashfs.so for writing generic inodes to a meta writer and another function to libsqfshelper.a that turns a tree node to an inode. That way, the tree serialization code can be expressed in terms of those functions and a bulk of the independend code can be moved over to libsquashfs.so Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-04Add fstree independend directory writer to libsquashfs.soDavid Oberhollenzer
This commit adds a directory writer to libsquashfs that wrapps a meta data writer and provides a higher-level interface for writing directory entries. Under the hood it enforces the rules that squashfs insists upon. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-02Move fstree independend parts of xattr_reader to libsquashfs.soDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-01Internalize the layout of the id_table_t structureDavid Oberhollenzer
As an opaque struct it has a chance to change its layout in the future without breaking ABI compatibiliy. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-01Move command line processing stuff out of compressor codeDavid Oberhollenzer
This commit moves stuff like printing help text, command line option processing and enumerating available processors on stdout out of the generic compressor code. The option string is replaced with a structure that directly exposese the tweakable parameters for all compressors. A function for parsing the command line arguments into this structure is added in sqfshelper. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-01API cleanup: Shuffle declarations aroundDavid Oberhollenzer
Move declarations for stuff that is defined in libsquashfs.so into the public headers and declarations for stuff that isn't, out of there. Also move the meta reader/writer helper functions to their respective headers. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-01Break up squashfs.h into topic related headersDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-01Install libsquashfs.so headers on the system in "sqfs" subdirectoryDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-09-01Move some application specific stuff out of libutilDavid Oberhollenzer
This commit does the following: - canonicalize_name is moved to libfstree - source_date_epoch is only used inside libfstree, so it's also moved over and can later be completely internalized - print_version is moved over to sqfshelper. Mainly so it doesn't end up in libsquashfs.so for no sane reason. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-31Turn libsquashfs.a into a shared libraryDavid Oberhollenzer
This of course entails turning the entire project over to libtool magic. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-31Split libsquashfs.a into low seperate librariesDavid Oberhollenzer
The idea is to make libsquashfs.a independend of libfstree.a, so it becomes a general purpose squashfs manipulation library. All the high level glue code for libfstree.a and utilites that are overly specific with to tools are moved to a seperate librarby. This commit makes the first step by moving the stuff with dependencies on libfstree to a seperate library. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-31Merge libcompress.a into libsquashfs.aDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-27Merge alloc_flex conditionals into oneDavid Oberhollenzer
It is shorter and less confusing for coverity. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-26Tune the paranoia down a bitDavid Oberhollenzer
size_t is guaranteed to be large enough to measure the size of things in memory, so when doing exactely that (e.g. strlen(a) + strlen(b)), checking for overflow is pointless since both objects are already in memory. If the addition would overflow, the two strings would occupy more memory than addressable. (Possible exception being some kind of harward style architecture with the two strings being in different kinds of memory of course.) Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-25Propperly set errno in read_inode_slink error pathDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-25Size accounting + alloc() overflow checking, round #3David Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-23Check against format limits in meta_reader_read_dir_headerDavid Oberhollenzer
The SquashFS kernel implementation insists that a directory header is followed by no more than an upper bound of entries, way less than what the filed itself actually supports. This commit makes sure that the meta_reader_read_dir_header function also enforces that same limit. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-23Size accounting + alloc() overflow checking, round #2David Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-23Do bounds checking in metadata readerDavid Oberhollenzer
In all cases where metadata blocks are read, we can roughly (in some cases even preciesly) say in what range those metadata blocks will be, so it makes sense to throw an error if an attempt is made to wander outside this range. Furthermore, when reading from an uncompressed block, it is more reasonable to check against the actual block bounds than to padd it with 0 bytes. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-23Some simple search/replace cases for allocationDavid Oberhollenzer
This commit exchanges some malloc(x + y * z) patterns that can be found with a simple git grep and are obvious for the new wrappers. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-23Add wrappers for calloc style functions with size overflow checkingDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-22deserialize_tree: filter out directory loopsDavid Oberhollenzer
The tree deserializer does a recursive depth-first search to populate the directory tree, moving back and forth between the directory listing containing the inode references and the inode table pointing to the list of child inodes. It is completely unaware of hard links and creates duplicate nodes instead. It is possible to create a malicious SquashFS image that contains a directory that contains as child a reference to its own inode. This can also be done transitively (i.e. directory contains its own parent or grand parent), leading to infinite recursion (actually finite, since it terminates once all stack memory is exhausted). This commit adds a simple check to see if a node has the same inode number as any of its would-be parents. If it does, the node is discarded and a warning message is emitted. Other cases with arbitrary layers of indirection could be constructed as well (e.g. dir 'a' contains hard link to 'b' and 'b' one back to 'a'), but the sub hierarchies are always expanded, this check should catch that too. Reported-by: Zachary Dremann <dremann@gmail.com> Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-21Fix "no attributes" sentinel value for xattr readerDavid Oberhollenzer
An inode can be of extended type for reasons other than having extended attributes and simply set the xattr ID to 0xFFFFFFFF to indicate that it doesn't have extended attributes. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-19Fix file list generation: break any pre-existing connectionDavid Oberhollenzer
If the linked list pointer was already used before, break up the connection so we don't risk running into a loop or something when regenerating the list. Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-19Minor improvements for parallel block processorDavid Oberhollenzer
- Fewer lock aquires in worker function - There is no point in locking/unlocking for inserting the completed block if we are going to lock again immediately in the next iteration -> Merge those two critical sections into one - Constant time queue insertion - Bypass queue entirely if there is nothing to do for a block Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
2019-08-19Fix memory leak in data writer fragment deduplicationDavid Oberhollenzer
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>