On 2017/2/28 7:40, Jaegeuk Kim wrote: > On 02/27, heyunlei wrote: >> Hi Jaegeuk, >> >> On 2017/2/27 9:59, Hou Pengyang wrote: >>> We use has_not_enough_free_secs to check if there are enough free segments, >>> >>> (free_sections(sbi) + freed) <= >>> (node_secs + 2 * dent_secs + imeta_secs + >>> reserved_sections(sbi) + needed); >> >> Now node SSR is enable, how can we change this condition simply as: >> >> (free_sections(sbi) + freed) <= reserved_sections(sbi) > > Currently, we can't guarantee successful SSR allocation all the time.
Okay, thanks. > >> If dirty node pages can find SSR segment, write it in SSR mode, >> >> or it will consume free segments, and then we do FG_GC. >> >> >> >> Another thing is that each process has a chance to do FG_GC, it's no problem >> >> for all normal processes, but may be not a good thing for a process within >> >> certain limits such as lower priority, with bound cpu to do FG_GC. Maybe it >> >> will block system for a long time. So what can we do to improve this? > > Could you please elaborate this in more detail? System bind lots of background processes to one or two cpus. In some corner case, thread need to do FG GC will compete with other background processes. FG GC can not be completed in time and then blocks other processes. Thanks. > > Thanks, > >> >> Thanks. >> > > . > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ Linux-f2fs-devel mailing list Linux-f2fs-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
