diff options
Diffstat (limited to 'hw/9pfs/9p.c')
-rw-r--r-- | hw/9pfs/9p.c | 124 |
1 files changed, 102 insertions, 22 deletions
diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c index af636cf..8b001b9 100644 --- a/hw/9pfs/9p.c +++ b/hw/9pfs/9p.c @@ -434,16 +434,24 @@ void coroutine_fn v9fs_reclaim_fd(V9fsPDU *pdu) V9fsFidState *f; GHashTableIter iter; gpointer fid; + int err; + int nclosed = 0; + + /* prevent multiple coroutines running this function simultaniously */ + if (s->reclaiming) { + return; + } + s->reclaiming = true; g_hash_table_iter_init(&iter, s->fids); QSLIST_HEAD(, V9fsFidState) reclaim_list = QSLIST_HEAD_INITIALIZER(reclaim_list); + /* Pick FIDs to be closed, collect them on reclaim_list. */ while (g_hash_table_iter_next(&iter, &fid, (gpointer *) &f)) { /* - * Unlink fids cannot be reclaimed. Check - * for them and skip them. Also skip fids + * Unlinked fids cannot be reclaimed, skip those, and also skip fids * currently being operated on. */ if (f->ref || f->flags & FID_NON_RECLAIMABLE) { @@ -493,23 +501,42 @@ void coroutine_fn v9fs_reclaim_fd(V9fsPDU *pdu) } } /* - * Now close the fid in reclaim list. Free them if they - * are already clunked. + * Close the picked FIDs altogether on a background I/O driver thread. Do + * this all at once to keep latency (i.e. amount of thread hops between main + * thread <-> fs driver background thread) as low as possible. */ + v9fs_co_run_in_worker({ + QSLIST_FOREACH(f, &reclaim_list, reclaim_next) { + err = (f->fid_type == P9_FID_DIR) ? + s->ops->closedir(&s->ctx, &f->fs_reclaim) : + s->ops->close(&s->ctx, &f->fs_reclaim); + + /* 'man 2 close' suggests to ignore close() errors except of EBADF */ + if (unlikely(err && errno == EBADF)) { + /* + * unexpected case as FIDs were picked above by having a valid + * file descriptor + */ + error_report("9pfs: v9fs_reclaim_fd() WARNING: close() failed with EBADF"); + } else { + /* total_open_fd must only be mutated on main thread */ + nclosed++; + } + } + }); + total_open_fd -= nclosed; + /* Free the closed FIDs. */ while (!QSLIST_EMPTY(&reclaim_list)) { f = QSLIST_FIRST(&reclaim_list); QSLIST_REMOVE(&reclaim_list, f, V9fsFidState, reclaim_next); - if (f->fid_type == P9_FID_FILE) { - v9fs_co_close(pdu, &f->fs_reclaim); - } else if (f->fid_type == P9_FID_DIR) { - v9fs_co_closedir(pdu, &f->fs_reclaim); - } /* * Now drop the fid reference, free it * if clunked. */ put_fid(pdu, f); } + + s->reclaiming = false; } /* @@ -1574,6 +1601,11 @@ out_nofid: pdu_complete(pdu, err); } +static bool fid_has_valid_file_handle(V9fsState *s, V9fsFidState *fidp) +{ + return s->ops->has_valid_file_handle(fidp->fid_type, &fidp->fs); +} + static void coroutine_fn v9fs_getattr(void *opaque) { int32_t fid; @@ -1596,11 +1628,11 @@ static void coroutine_fn v9fs_getattr(void *opaque) retval = -ENOENT; goto out_nofid; } - /* - * Currently we only support BASIC fields in stat, so there is no - * need to look at request_mask. - */ - retval = v9fs_co_lstat(pdu, &fidp->path, &stbuf); + if (fid_has_valid_file_handle(pdu->s, fidp)) { + retval = v9fs_co_fstat(pdu, fidp, &stbuf); + } else { + retval = v9fs_co_lstat(pdu, &fidp->path, &stbuf); + } if (retval < 0) { goto out; } @@ -1703,7 +1735,11 @@ static void coroutine_fn v9fs_setattr(void *opaque) } else { times[1].tv_nsec = UTIME_OMIT; } - err = v9fs_co_utimensat(pdu, &fidp->path, times); + if (fid_has_valid_file_handle(pdu->s, fidp)) { + err = v9fs_co_futimens(pdu, fidp, times); + } else { + err = v9fs_co_utimensat(pdu, &fidp->path, times); + } if (err < 0) { goto out; } @@ -1728,7 +1764,11 @@ static void coroutine_fn v9fs_setattr(void *opaque) } } if (v9iattr.valid & (P9_ATTR_SIZE)) { - err = v9fs_co_truncate(pdu, &fidp->path, v9iattr.size); + if (fid_has_valid_file_handle(pdu->s, fidp)) { + err = v9fs_co_ftruncate(pdu, fidp, v9iattr.size); + } else { + err = v9fs_co_truncate(pdu, &fidp->path, v9iattr.size); + } if (err < 0) { goto out; } @@ -1772,6 +1812,21 @@ static bool same_stat_id(const struct stat *a, const struct stat *b) return a->st_dev == b->st_dev && a->st_ino == b->st_ino; } +/* + * Returns a (newly allocated) comma-separated string presentation of the + * passed array for logging (tracing) purpose for trace event "v9fs_walk". + * + * It is caller's responsibility to free the returned string. + */ +static char *trace_v9fs_walk_wnames(V9fsString *wnames, size_t nwnames) +{ + g_autofree char **arr = g_malloc0_n(nwnames + 1, sizeof(char *)); + for (size_t i = 0; i < nwnames; ++i) { + arr[i] = wnames[i].data; + } + return g_strjoinv(", ", arr); +} + static void coroutine_fn v9fs_walk(void *opaque) { int name_idx, nwalked; @@ -1785,6 +1840,7 @@ static void coroutine_fn v9fs_walk(void *opaque) size_t offset = 7; int32_t fid, newfid; P9ARRAY_REF(V9fsString) wnames = NULL; + g_autofree char *trace_wnames = NULL; V9fsFidState *fidp; V9fsFidState *newfidp = NULL; V9fsPDU *pdu = opaque; @@ -1798,11 +1854,9 @@ static void coroutine_fn v9fs_walk(void *opaque) } offset += err; - trace_v9fs_walk(pdu->tag, pdu->id, fid, newfid, nwnames); - if (nwnames > P9_MAXWELEM) { err = -EINVAL; - goto out_nofid; + goto out_nofid_nownames; } if (nwnames) { P9ARRAY_NEW(V9fsString, wnames, nwnames); @@ -1812,15 +1866,23 @@ static void coroutine_fn v9fs_walk(void *opaque) for (i = 0; i < nwnames; i++) { err = pdu_unmarshal(pdu, offset, "s", &wnames[i]); if (err < 0) { - goto out_nofid; + goto out_nofid_nownames; } if (name_is_illegal(wnames[i].data)) { err = -ENOENT; - goto out_nofid; + goto out_nofid_nownames; } offset += err; } + if (trace_event_get_state_backends(TRACE_V9FS_WALK)) { + trace_wnames = trace_v9fs_walk_wnames(wnames, nwnames); + trace_v9fs_walk(pdu->tag, pdu->id, fid, newfid, nwnames, + trace_wnames); + } + } else { + trace_v9fs_walk(pdu->tag, pdu->id, fid, newfid, nwnames, ""); } + fidp = get_fid(pdu, fid); if (fidp == NULL) { err = -ENOENT; @@ -1955,7 +2017,11 @@ out: } v9fs_path_free(&dpath); v9fs_path_free(&path); + goto out_pdu_complete; +out_nofid_nownames: + trace_v9fs_walk(pdu->tag, pdu->id, fid, newfid, nwnames, "<?>"); out_nofid: +out_pdu_complete: pdu_complete(pdu, err); } @@ -1980,6 +2046,7 @@ static void coroutine_fn v9fs_open(void *opaque) V9fsFidState *fidp; V9fsPDU *pdu = opaque; V9fsState *s = pdu->s; + g_autofree char *trace_oflags = NULL; if (s->proto_version == V9FS_PROTO_2000L) { err = pdu_unmarshal(pdu, offset, "dd", &fid, &mode); @@ -1991,7 +2058,13 @@ static void coroutine_fn v9fs_open(void *opaque) if (err < 0) { goto out_nofid; } - trace_v9fs_open(pdu->tag, pdu->id, fid, mode); + if (trace_event_get_state_backends(TRACE_V9FS_OPEN)) { + trace_oflags = qemu_open_flags_tostr( + (s->proto_version == V9FS_PROTO_2000L) ? + dotl_to_open_flags(mode) : omode_to_uflags(mode) + ); + trace_v9fs_open(pdu->tag, pdu->id, fid, mode, trace_oflags); + } fidp = get_fid(pdu, fid); if (fidp == NULL) { @@ -2587,6 +2660,11 @@ static void coroutine_fn v9fs_readdir(void *opaque) retval = -EINVAL; goto out_nofid; } + if (fidp->fid_type != P9_FID_DIR) { + warn_report_once("9p: bad client: T_readdir on non-directory stream"); + retval = -ENOTDIR; + goto out; + } if (!fidp->fs.dir.stream) { retval = -EINVAL; goto out; @@ -4284,6 +4362,8 @@ int v9fs_device_realize_common(V9fsState *s, const V9fsTransport *t, s->ctx.fst = &fse->fst; fsdev_throttle_init(s->ctx.fst); + s->reclaiming = false; + rc = 0; out: if (rc) { |