----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.csiden.org/r/166/#review550 -----------------------------------------------------------
Ship it! Ship It! - Josef 'Jeff' Sipek On March 5, 2015, 12:24 a.m., Justin Gibbs wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.csiden.org/r/166/ > ----------------------------------------------------------- > > (Updated March 5, 2015, 12:24 a.m.) > > > Review request for OpenZFS Developer Mailing List. > > > Bugs: 5661 > https://www.illumos.org/projects/illumos-gate//issues/5661 > > > Repository: illumos-gate > > > Description > ------- > > Make lz4 the default compression algrithm for datasets with > > the "compression=on" property setting. lzjb will be used if the > > pool does not have "feature@lz4_compress" enabled. > > uts/common/fs/zfs/sys/zio.h: > Replace ZIO_COMPRESS_ON_VALUE with > ZIO_COMPRESS_LEGACY_ON_VALUE and ZIO_COMPRESS_LZ4_ON_VALUE. > The former is used for pools without "feature@lz4_compress" > enabled. > > Allow ZIO_COMPRESS_ON on bootable datasets without > restriction to certain values of (the now removed) > ZIO_COMPRESS_ON_VALUE constant. The boot loader > knows how to decompress both lzjb and lz4, the two > types of compression that ZIO_COMPRESS_ON can mean. > The loader also does not write data, and existing > blkptrs have the compression type explicitly listed > (ZIO_COMPRESS_ON never hits the media), making > ZIO_COMPRESS_ON bootfs safe. > > uts/common/fs/zfs/dmu.c: > uts/common/fs/zfs/dmu_objset.c: > uts/common/fs/zfs/sys/zio.h: > uts/common/fs/zfs/zio_compress.c: > Pass the spa into zio_compress_select() so that > enabled spa features can be considered with selecting > a compression algorithm. > > uts/common/fs/zfs/dmu_objset.c: > uts/common/fs/zfs/zio_compress.c: > Historically, callers of zio_compress_select() > converted a parent value of ZIO_COMPRESS_ON to > ZIO_COMPRESS_ON_VALUE (on -> lzjb). This made > it impossible for zio_compress_select() to > differentiate between a user explicitly requesting > lzjb or the default compression algorithm. Allow > the parent/inherited value to be expressed as > ZIO_COMPRESS_ON and convert it to the appropriate > default compression algorithm within > zio_compress_select(). > > uts/common/fs/zfs/dmu.c: > In dmu_write_policy(), rather than duplicate logic > which is now in zio_compress_select(), use > zio_compress_select() to determine the correct > compression algorithm to use for metadata. > > uts/common/fs/zfs/dmu_objset.c: > Allow the MOS to use the default compression algorithm > rather than hard code lzjb. > > > Diffs > ----- > > usr/src/uts/common/fs/zfs/zio_compress.c > 0d77f246c4a755cef5bbdc1e9cd3b34882b01730 > usr/src/uts/common/fs/zfs/sys/zio.h > 71bb6b96a01c51437cd5c6622e404531b5e8de17 > usr/src/uts/common/fs/zfs/dmu_objset.c > 8593e2c94f6cb5e5b3866f5763f6230e25294716 > usr/src/uts/common/fs/zfs/dmu.c c541e04d54297243c9d6a342b7a8df742e8d7ed2 > usr/src/man/man1m/zfs.1m b679b77ba4e58fbeaf7a63052e49398ca2c80212 > > Diff: https://reviews.csiden.org/r/166/diff/ > > > Testing > ------- > > ztest via zloop on illumos. ZFS test suite on FreeBSD and illumos. > > > Thanks, > > Justin Gibbs > >
_______________________________________________ developer mailing list [email protected] http://lists.open-zfs.org/mailman/listinfo/developer
