From patchwork Thu Aug 9 14:33:55 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Dominique Martinet X-Patchwork-Id: 10561477 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id ADF6A96FA for ; Thu, 9 Aug 2018 14:34:26 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 9B1172B4A6 for ; Thu, 9 Aug 2018 14:34:26 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 8EB742B55C; Thu, 9 Aug 2018 14:34:26 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 9423E2B4A6 for ; Thu, 9 Aug 2018 14:34:24 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1732163AbeHIQ71 (ORCPT ); Thu, 9 Aug 2018 12:59:27 -0400 Received: from nautica.notk.org ([91.121.71.147]:37662 "EHLO nautica.notk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1730453AbeHIQ71 (ORCPT ); Thu, 9 Aug 2018 12:59:27 -0400 Received: by nautica.notk.org (Postfix, from userid 1001) id C2739C009; Thu, 9 Aug 2018 16:34:08 +0200 (CEST) From: Dominique Martinet To: v9fs-developer@lists.sourceforge.net Cc: Dominique Martinet , linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, Greg Kurz , Matthew Wilcox , Jun Piao Subject: [PATCH v3 1/2] net/9p: embed fcall in req to round down buffer allocs Date: Thu, 9 Aug 2018 16:33:55 +0200 Message-Id: <1533825236-22896-1-git-send-email-asmadeus@codewreck.org> X-Mailer: git-send-email 1.7.10.4 In-Reply-To: <1533177452-2165-1-git-send-email-asmadeus@codewreck.org> References: <1533177452-2165-1-git-send-email-asmadeus@codewreck.org> Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP From: Dominique Martinet 'msize' is often a power of two, or at least page-aligned, so avoiding an overhead of two dozen bytes for each allocation will help the allocator do its work and reduce memory fragmentation. Suggested-by: Matthew Wilcox Signed-off-by: Dominique Martinet Reviewed-by: Greg Kurz Cc: Matthew Wilcox Cc: Jun Piao Acked-by: Jun Piao --- v2: - Add extra label to not free uninitialized memory on alloc failure - Rename p9_fcall_alloc to 9p_fcall_init - Add a p9_fcall_fini function to echo to init v3: - use p9_call_fini() in rdma_request() instead of kfree, the code was in the following patch in previous version include/net/9p/client.h | 5 +- net/9p/client.c | 167 +++++++++++++++++++++------------------- net/9p/trans_fd.c | 12 +-- net/9p/trans_rdma.c | 29 +++---- net/9p/trans_virtio.c | 18 ++--- net/9p/trans_xen.c | 12 +-- 6 files changed, 125 insertions(+), 118 deletions(-) diff --git a/include/net/9p/client.h b/include/net/9p/client.h index a4dc42c53d18..c2671d40bb6b 100644 --- a/include/net/9p/client.h +++ b/include/net/9p/client.h @@ -95,8 +95,8 @@ struct p9_req_t { int status; int t_err; wait_queue_head_t wq; - struct p9_fcall *tc; - struct p9_fcall *rc; + struct p9_fcall tc; + struct p9_fcall rc; void *aux; struct list_head req_list; }; @@ -230,6 +230,7 @@ int p9_client_mkdir_dotl(struct p9_fid *fid, const char *name, int mode, kgid_t gid, struct p9_qid *); int p9_client_lock_dotl(struct p9_fid *fid, struct p9_flock *flock, u8 *status); int p9_client_getlock_dotl(struct p9_fid *fid, struct p9_getlock *fl); +void p9_fcall_fini(struct p9_fcall *fc); struct p9_req_t *p9_tag_lookup(struct p9_client *, u16); void p9_client_cb(struct p9_client *c, struct p9_req_t *req, int status); diff --git a/net/9p/client.c b/net/9p/client.c index 88db45966740..ed78751aee7c 100644 --- a/net/9p/client.c +++ b/net/9p/client.c @@ -231,16 +231,20 @@ static int parse_opts(char *opts, struct p9_client *clnt) return ret; } -static struct p9_fcall *p9_fcall_alloc(int alloc_msize) +static int p9_fcall_init(struct p9_fcall *fc, int alloc_msize) { - struct p9_fcall *fc; - fc = kmalloc(sizeof(struct p9_fcall) + alloc_msize, GFP_NOFS); - if (!fc) - return NULL; + fc->sdata = kmalloc(alloc_msize, GFP_NOFS); + if (!fc->sdata) + return -ENOMEM; fc->capacity = alloc_msize; - fc->sdata = (char *) fc + sizeof(struct p9_fcall); - return fc; + return 0; +} + +void p9_fcall_fini(struct p9_fcall *fc) +{ + kfree(fc->sdata); } +EXPORT_SYMBOL(p9_fcall_fini); static struct kmem_cache *p9_req_cache; @@ -263,13 +267,13 @@ p9_tag_alloc(struct p9_client *c, int8_t type, unsigned int max_size) if (!req) return NULL; - req->tc = p9_fcall_alloc(alloc_msize); - req->rc = p9_fcall_alloc(alloc_msize); - if (!req->tc || !req->rc) + if (p9_fcall_init(&req->tc, alloc_msize)) + goto free_req; + if (p9_fcall_init(&req->rc, alloc_msize)) goto free; - p9pdu_reset(req->tc); - p9pdu_reset(req->rc); + p9pdu_reset(&req->tc); + p9pdu_reset(&req->rc); req->status = REQ_STATUS_ALLOC; init_waitqueue_head(&req->wq); INIT_LIST_HEAD(&req->req_list); @@ -281,7 +285,7 @@ p9_tag_alloc(struct p9_client *c, int8_t type, unsigned int max_size) GFP_NOWAIT); else tag = idr_alloc(&c->reqs, req, 0, P9_NOTAG, GFP_NOWAIT); - req->tc->tag = tag; + req->tc.tag = tag; spin_unlock_irq(&c->lock); idr_preload_end(); if (tag < 0) @@ -290,8 +294,9 @@ p9_tag_alloc(struct p9_client *c, int8_t type, unsigned int max_size) return req; free: - kfree(req->tc); - kfree(req->rc); + p9_fcall_fini(&req->tc); + p9_fcall_fini(&req->rc); +free_req: kmem_cache_free(p9_req_cache, req); return ERR_PTR(-ENOMEM); } @@ -329,14 +334,14 @@ EXPORT_SYMBOL(p9_tag_lookup); static void p9_free_req(struct p9_client *c, struct p9_req_t *r) { unsigned long flags; - u16 tag = r->tc->tag; + u16 tag = r->tc.tag; p9_debug(P9_DEBUG_MUX, "clnt %p req %p tag: %d\n", c, r, tag); spin_lock_irqsave(&c->lock, flags); idr_remove(&c->reqs, tag); spin_unlock_irqrestore(&c->lock, flags); - kfree(r->tc); - kfree(r->rc); + p9_fcall_fini(&r->tc); + p9_fcall_fini(&r->rc); kmem_cache_free(p9_req_cache, r); } @@ -368,7 +373,7 @@ static void p9_tag_cleanup(struct p9_client *c) */ void p9_client_cb(struct p9_client *c, struct p9_req_t *req, int status) { - p9_debug(P9_DEBUG_MUX, " tag %d\n", req->tc->tag); + p9_debug(P9_DEBUG_MUX, " tag %d\n", req->tc.tag); /* * This barrier is needed to make sure any change made to req before @@ -378,7 +383,7 @@ void p9_client_cb(struct p9_client *c, struct p9_req_t *req, int status) req->status = status; wake_up(&req->wq); - p9_debug(P9_DEBUG_MUX, "wakeup: %d\n", req->tc->tag); + p9_debug(P9_DEBUG_MUX, "wakeup: %d\n", req->tc.tag); } EXPORT_SYMBOL(p9_client_cb); @@ -449,18 +454,18 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) int err; int ecode; - err = p9_parse_header(req->rc, NULL, &type, NULL, 0); - if (req->rc->size >= c->msize) { + err = p9_parse_header(&req->rc, NULL, &type, NULL, 0); + if (req->rc.size >= c->msize) { p9_debug(P9_DEBUG_ERROR, "requested packet size too big: %d\n", - req->rc->size); + req->rc.size); return -EIO; } /* * dump the response from server * This should be after check errors which poplulate pdu_fcall. */ - trace_9p_protocol_dump(c, req->rc); + trace_9p_protocol_dump(c, &req->rc); if (err) { p9_debug(P9_DEBUG_ERROR, "couldn't parse header %d\n", err); return err; @@ -470,7 +475,7 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) if (!p9_is_proto_dotl(c)) { char *ename; - err = p9pdu_readf(req->rc, c->proto_version, "s?d", + err = p9pdu_readf(&req->rc, c->proto_version, "s?d", &ename, &ecode); if (err) goto out_err; @@ -486,7 +491,7 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) } kfree(ename); } else { - err = p9pdu_readf(req->rc, c->proto_version, "d", &ecode); + err = p9pdu_readf(&req->rc, c->proto_version, "d", &ecode); err = -ecode; p9_debug(P9_DEBUG_9P, "<<< RLERROR (%d)\n", -ecode); @@ -520,12 +525,12 @@ static int p9_check_zc_errors(struct p9_client *c, struct p9_req_t *req, int8_t type; char *ename = NULL; - err = p9_parse_header(req->rc, NULL, &type, NULL, 0); + err = p9_parse_header(&req->rc, NULL, &type, NULL, 0); /* * dump the response from server * This should be after parse_header which poplulate pdu_fcall. */ - trace_9p_protocol_dump(c, req->rc); + trace_9p_protocol_dump(c, &req->rc); if (err) { p9_debug(P9_DEBUG_ERROR, "couldn't parse header %d\n", err); return err; @@ -540,13 +545,13 @@ static int p9_check_zc_errors(struct p9_client *c, struct p9_req_t *req, /* 7 = header size for RERROR; */ int inline_len = in_hdrlen - 7; - len = req->rc->size - req->rc->offset; + len = req->rc.size - req->rc.offset; if (len > (P9_ZC_HDR_SZ - 7)) { err = -EFAULT; goto out_err; } - ename = &req->rc->sdata[req->rc->offset]; + ename = &req->rc.sdata[req->rc.offset]; if (len > inline_len) { /* We have error in external buffer */ if (!copy_from_iter_full(ename + inline_len, @@ -556,7 +561,7 @@ static int p9_check_zc_errors(struct p9_client *c, struct p9_req_t *req, } } ename = NULL; - err = p9pdu_readf(req->rc, c->proto_version, "s?d", + err = p9pdu_readf(&req->rc, c->proto_version, "s?d", &ename, &ecode); if (err) goto out_err; @@ -572,7 +577,7 @@ static int p9_check_zc_errors(struct p9_client *c, struct p9_req_t *req, } kfree(ename); } else { - err = p9pdu_readf(req->rc, c->proto_version, "d", &ecode); + err = p9pdu_readf(&req->rc, c->proto_version, "d", &ecode); err = -ecode; p9_debug(P9_DEBUG_9P, "<<< RLERROR (%d)\n", -ecode); @@ -605,7 +610,7 @@ static int p9_client_flush(struct p9_client *c, struct p9_req_t *oldreq) int16_t oldtag; int err; - err = p9_parse_header(oldreq->tc, NULL, NULL, &oldtag, 1); + err = p9_parse_header(&oldreq->tc, NULL, NULL, &oldtag, 1); if (err) return err; @@ -649,12 +654,12 @@ static struct p9_req_t *p9_client_prepare_req(struct p9_client *c, return req; /* marshall the data */ - p9pdu_prepare(req->tc, req->tc->tag, type); - err = p9pdu_vwritef(req->tc, c->proto_version, fmt, ap); + p9pdu_prepare(&req->tc, req->tc.tag, type); + err = p9pdu_vwritef(&req->tc, c->proto_version, fmt, ap); if (err) goto reterr; - p9pdu_finalize(c, req->tc); - trace_9p_client_req(c, type, req->tc->tag); + p9pdu_finalize(c, &req->tc); + trace_9p_client_req(c, type, req->tc.tag); return req; reterr: p9_free_req(c, req); @@ -739,7 +744,7 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...) goto reterr; err = p9_check_errors(c, req); - trace_9p_client_res(c, type, req->rc->tag, err); + trace_9p_client_res(c, type, req->rc.tag, err); if (!err) return req; reterr: @@ -821,7 +826,7 @@ static struct p9_req_t *p9_client_zc_rpc(struct p9_client *c, int8_t type, goto reterr; err = p9_check_zc_errors(c, req, uidata, in_hdrlen); - trace_9p_client_res(c, type, req->rc->tag, err); + trace_9p_client_res(c, type, req->rc.tag, err); if (!err) return req; reterr: @@ -904,10 +909,10 @@ static int p9_client_version(struct p9_client *c) if (IS_ERR(req)) return PTR_ERR(req); - err = p9pdu_readf(req->rc, c->proto_version, "ds", &msize, &version); + err = p9pdu_readf(&req->rc, c->proto_version, "ds", &msize, &version); if (err) { p9_debug(P9_DEBUG_9P, "version error %d\n", err); - trace_9p_protocol_dump(c, req->rc); + trace_9p_protocol_dump(c, &req->rc); goto error; } @@ -1056,9 +1061,9 @@ struct p9_fid *p9_client_attach(struct p9_client *clnt, struct p9_fid *afid, goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "Q", &qid); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Q", &qid); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); goto error; } @@ -1113,9 +1118,9 @@ struct p9_fid *p9_client_walk(struct p9_fid *oldfid, uint16_t nwname, goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "R", &nwqids, &wqids); + err = p9pdu_readf(&req->rc, clnt->proto_version, "R", &nwqids, &wqids); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); goto clunk_fid; } @@ -1180,9 +1185,9 @@ int p9_client_open(struct p9_fid *fid, int mode) goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "Qd", &qid, &iounit); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Qd", &qid, &iounit); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto free_and_error; } @@ -1224,9 +1229,9 @@ int p9_client_create_dotl(struct p9_fid *ofid, const char *name, u32 flags, u32 goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "Qd", qid, &iounit); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Qd", qid, &iounit); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto free_and_error; } @@ -1269,9 +1274,9 @@ int p9_client_fcreate(struct p9_fid *fid, const char *name, u32 perm, int mode, goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "Qd", &qid, &iounit); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Qd", &qid, &iounit); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto free_and_error; } @@ -1308,9 +1313,9 @@ int p9_client_symlink(struct p9_fid *dfid, const char *name, goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "Q", qid); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Q", qid); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto free_and_error; } @@ -1506,10 +1511,10 @@ p9_client_read(struct p9_fid *fid, u64 offset, struct iov_iter *to, int *err) break; } - *err = p9pdu_readf(req->rc, clnt->proto_version, + *err = p9pdu_readf(&req->rc, clnt->proto_version, "D", &count, &dataptr); if (*err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); break; } @@ -1579,9 +1584,9 @@ p9_client_write(struct p9_fid *fid, u64 offset, struct iov_iter *from, int *err) break; } - *err = p9pdu_readf(req->rc, clnt->proto_version, "d", &count); + *err = p9pdu_readf(&req->rc, clnt->proto_version, "d", &count); if (*err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); break; } @@ -1623,9 +1628,9 @@ struct p9_wstat *p9_client_stat(struct p9_fid *fid) goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "wS", &ignored, ret); + err = p9pdu_readf(&req->rc, clnt->proto_version, "wS", &ignored, ret); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); goto error; } @@ -1676,9 +1681,9 @@ struct p9_stat_dotl *p9_client_getattr_dotl(struct p9_fid *fid, goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "A", ret); + err = p9pdu_readf(&req->rc, clnt->proto_version, "A", ret); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); goto error; } @@ -1828,11 +1833,11 @@ int p9_client_statfs(struct p9_fid *fid, struct p9_rstatfs *sb) goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "ddqqqqqqd", &sb->type, - &sb->bsize, &sb->blocks, &sb->bfree, &sb->bavail, - &sb->files, &sb->ffree, &sb->fsid, &sb->namelen); + err = p9pdu_readf(&req->rc, clnt->proto_version, "ddqqqqqqd", &sb->type, + &sb->bsize, &sb->blocks, &sb->bfree, &sb->bavail, + &sb->files, &sb->ffree, &sb->fsid, &sb->namelen); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); goto error; } @@ -1936,9 +1941,9 @@ struct p9_fid *p9_client_xattrwalk(struct p9_fid *file_fid, err = PTR_ERR(req); goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "q", attr_size); + err = p9pdu_readf(&req->rc, clnt->proto_version, "q", attr_size); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); p9_free_req(clnt, req); goto clunk_fid; } @@ -2024,9 +2029,9 @@ int p9_client_readdir(struct p9_fid *fid, char *data, u32 count, u64 offset) goto error; } - err = p9pdu_readf(req->rc, clnt->proto_version, "D", &count, &dataptr); + err = p9pdu_readf(&req->rc, clnt->proto_version, "D", &count, &dataptr); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto free_and_error; } if (rsize < count) { @@ -2065,9 +2070,9 @@ int p9_client_mknod_dotl(struct p9_fid *fid, const char *name, int mode, if (IS_ERR(req)) return PTR_ERR(req); - err = p9pdu_readf(req->rc, clnt->proto_version, "Q", qid); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Q", qid); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto error; } p9_debug(P9_DEBUG_9P, "<<< RMKNOD qid %x.%llx.%x\n", qid->type, @@ -2096,9 +2101,9 @@ int p9_client_mkdir_dotl(struct p9_fid *fid, const char *name, int mode, if (IS_ERR(req)) return PTR_ERR(req); - err = p9pdu_readf(req->rc, clnt->proto_version, "Q", qid); + err = p9pdu_readf(&req->rc, clnt->proto_version, "Q", qid); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto error; } p9_debug(P9_DEBUG_9P, "<<< RMKDIR qid %x.%llx.%x\n", qid->type, @@ -2131,9 +2136,9 @@ int p9_client_lock_dotl(struct p9_fid *fid, struct p9_flock *flock, u8 *status) if (IS_ERR(req)) return PTR_ERR(req); - err = p9pdu_readf(req->rc, clnt->proto_version, "b", status); + err = p9pdu_readf(&req->rc, clnt->proto_version, "b", status); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto error; } p9_debug(P9_DEBUG_9P, "<<< RLOCK status %i\n", *status); @@ -2162,11 +2167,11 @@ int p9_client_getlock_dotl(struct p9_fid *fid, struct p9_getlock *glock) if (IS_ERR(req)) return PTR_ERR(req); - err = p9pdu_readf(req->rc, clnt->proto_version, "bqqds", &glock->type, - &glock->start, &glock->length, &glock->proc_id, - &glock->client_id); + err = p9pdu_readf(&req->rc, clnt->proto_version, "bqqds", &glock->type, + &glock->start, &glock->length, &glock->proc_id, + &glock->client_id); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto error; } p9_debug(P9_DEBUG_9P, "<<< RGETLOCK type %i start %lld length %lld " @@ -2192,9 +2197,9 @@ int p9_client_readlink(struct p9_fid *fid, char **target) if (IS_ERR(req)) return PTR_ERR(req); - err = p9pdu_readf(req->rc, clnt->proto_version, "s", target); + err = p9pdu_readf(&req->rc, clnt->proto_version, "s", target); if (err) { - trace_9p_protocol_dump(clnt, req->rc); + trace_9p_protocol_dump(clnt, &req->rc); goto error; } p9_debug(P9_DEBUG_9P, "<<< RREADLINK target %s\n", *target); diff --git a/net/9p/trans_fd.c b/net/9p/trans_fd.c index e2ef3c782c53..20f46f13fe83 100644 --- a/net/9p/trans_fd.c +++ b/net/9p/trans_fd.c @@ -354,7 +354,7 @@ static void p9_read_work(struct work_struct *work) goto error; } - if (m->req->rc == NULL) { + if (m->req->rc.sdata == NULL) { p9_debug(P9_DEBUG_ERROR, "No recv fcall for tag %d (req %p), disconnecting!\n", m->rc.tag, m->req); @@ -362,7 +362,7 @@ static void p9_read_work(struct work_struct *work) err = -EIO; goto error; } - m->rc.sdata = (char *)m->req->rc + sizeof(struct p9_fcall); + m->rc.sdata = m->req->rc.sdata; memcpy(m->rc.sdata, m->tmp_buf, m->rc.capacity); m->rc.capacity = m->rc.size; } @@ -372,7 +372,7 @@ static void p9_read_work(struct work_struct *work) */ if ((m->req) && (m->rc.offset == m->rc.capacity)) { p9_debug(P9_DEBUG_TRANS, "got new packet\n"); - m->req->rc->size = m->rc.offset; + m->req->rc.size = m->rc.offset; spin_lock(&m->client->lock); if (m->req->status != REQ_STATUS_ERROR) status = REQ_STATUS_RCVD; @@ -469,8 +469,8 @@ static void p9_write_work(struct work_struct *work) p9_debug(P9_DEBUG_TRANS, "move req %p\n", req); list_move_tail(&req->req_list, &m->req_list); - m->wbuf = req->tc->sdata; - m->wsize = req->tc->size; + m->wbuf = req->tc.sdata; + m->wsize = req->tc.size; m->wpos = 0; spin_unlock(&m->client->lock); } @@ -663,7 +663,7 @@ static int p9_fd_request(struct p9_client *client, struct p9_req_t *req) struct p9_conn *m = &ts->conn; p9_debug(P9_DEBUG_TRANS, "mux %p task %p tcall %p id %d\n", - m, current, req->tc, req->tc->id); + m, current, &req->tc, req->tc.id); if (m->err < 0) return m->err; diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c index 2ab4574183c9..c60655c90c9e 100644 --- a/net/9p/trans_rdma.c +++ b/net/9p/trans_rdma.c @@ -122,7 +122,7 @@ struct p9_rdma_context { dma_addr_t busa; union { struct p9_req_t *req; - struct p9_fcall *rc; + struct p9_fcall rc; }; }; @@ -320,8 +320,8 @@ recv_done(struct ib_cq *cq, struct ib_wc *wc) if (wc->status != IB_WC_SUCCESS) goto err_out; - c->rc->size = wc->byte_len; - err = p9_parse_header(c->rc, NULL, NULL, &tag, 1); + c->rc.size = wc->byte_len; + err = p9_parse_header(&c->rc, NULL, NULL, &tag, 1); if (err) goto err_out; @@ -331,12 +331,13 @@ recv_done(struct ib_cq *cq, struct ib_wc *wc) /* Check that we have not yet received a reply for this request. */ - if (unlikely(req->rc)) { + if (unlikely(req->rc.sdata)) { pr_err("Duplicate reply for request %d", tag); goto err_out; } - req->rc = c->rc; + req->rc.size = c->rc.size; + req->rc.sdata = c->rc.sdata; p9_client_cb(client, req, REQ_STATUS_RCVD); out: @@ -361,7 +362,7 @@ send_done(struct ib_cq *cq, struct ib_wc *wc) container_of(wc->wr_cqe, struct p9_rdma_context, cqe); ib_dma_unmap_single(rdma->cm_id->device, - c->busa, c->req->tc->size, + c->busa, c->req->tc.size, DMA_TO_DEVICE); up(&rdma->sq_sem); kfree(c); @@ -401,7 +402,7 @@ post_recv(struct p9_client *client, struct p9_rdma_context *c) struct ib_sge sge; c->busa = ib_dma_map_single(rdma->cm_id->device, - c->rc->sdata, client->msize, + c->rc.sdata, client->msize, DMA_FROM_DEVICE); if (ib_dma_mapping_error(rdma->cm_id->device, c->busa)) goto error; @@ -443,9 +444,9 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) **/ if (unlikely(atomic_read(&rdma->excess_rc) > 0)) { if ((atomic_sub_return(1, &rdma->excess_rc) >= 0)) { - /* Got one ! */ - kfree(req->rc); - req->rc = NULL; + /* Got one! */ + p9_fcall_fini(&req->rc); + req->rc.sdata = NULL; goto dont_need_post_recv; } else { /* We raced and lost. */ @@ -459,7 +460,7 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) err = -ENOMEM; goto recv_error; } - rpl_context->rc = req->rc; + rpl_context->rc.sdata = req->rc.sdata; /* * Post a receive buffer for this request. We need to ensure @@ -479,7 +480,7 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) goto recv_error; } /* remove posted receive buffer from request structure */ - req->rc = NULL; + req->rc.sdata = NULL; dont_need_post_recv: /* Post the request */ @@ -491,7 +492,7 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) c->req = req; c->busa = ib_dma_map_single(rdma->cm_id->device, - c->req->tc->sdata, c->req->tc->size, + c->req->tc.sdata, c->req->tc.size, DMA_TO_DEVICE); if (ib_dma_mapping_error(rdma->cm_id->device, c->busa)) { err = -EIO; @@ -501,7 +502,7 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) c->cqe.done = send_done; sge.addr = c->busa; - sge.length = c->req->tc->size; + sge.length = c->req->tc.size; sge.lkey = rdma->pd->local_dma_lkey; wr.next = NULL; diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c index 7728b0acde09..3dd6ce1c0f2d 100644 --- a/net/9p/trans_virtio.c +++ b/net/9p/trans_virtio.c @@ -155,7 +155,7 @@ static void req_done(struct virtqueue *vq) } if (len) { - req->rc->size = len; + req->rc.size = len; p9_client_cb(chan->client, req, REQ_STATUS_RCVD); } } @@ -273,12 +273,12 @@ p9_virtio_request(struct p9_client *client, struct p9_req_t *req) out_sgs = in_sgs = 0; /* Handle out VirtIO ring buffers */ out = pack_sg_list(chan->sg, 0, - VIRTQUEUE_NUM, req->tc->sdata, req->tc->size); + VIRTQUEUE_NUM, req->tc.sdata, req->tc.size); if (out) sgs[out_sgs++] = chan->sg; in = pack_sg_list(chan->sg, out, - VIRTQUEUE_NUM, req->rc->sdata, req->rc->capacity); + VIRTQUEUE_NUM, req->rc.sdata, req->rc.capacity); if (in) sgs[out_sgs + in_sgs++] = chan->sg + out; @@ -416,15 +416,15 @@ p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req, out_nr_pages = DIV_ROUND_UP(n + offs, PAGE_SIZE); if (n != outlen) { __le32 v = cpu_to_le32(n); - memcpy(&req->tc->sdata[req->tc->size - 4], &v, 4); + memcpy(&req->tc.sdata[req->tc.size - 4], &v, 4); outlen = n; } /* The size field of the message must include the length of the * header and the length of the data. We didn't actually know * the length of the data until this point so add it in now. */ - sz = cpu_to_le32(req->tc->size + outlen); - memcpy(&req->tc->sdata[0], &sz, sizeof(sz)); + sz = cpu_to_le32(req->tc.size + outlen); + memcpy(&req->tc.sdata[0], &sz, sizeof(sz)); } else if (uidata) { int n = p9_get_mapped_pages(chan, &in_pages, uidata, inlen, &offs, &need_drop); @@ -433,7 +433,7 @@ p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req, in_nr_pages = DIV_ROUND_UP(n + offs, PAGE_SIZE); if (n != inlen) { __le32 v = cpu_to_le32(n); - memcpy(&req->tc->sdata[req->tc->size - 4], &v, 4); + memcpy(&req->tc.sdata[req->tc.size - 4], &v, 4); inlen = n; } } @@ -445,7 +445,7 @@ p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req, /* out data */ out = pack_sg_list(chan->sg, 0, - VIRTQUEUE_NUM, req->tc->sdata, req->tc->size); + VIRTQUEUE_NUM, req->tc.sdata, req->tc.size); if (out) sgs[out_sgs++] = chan->sg; @@ -464,7 +464,7 @@ p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req, * alloced memory and payload onto the user buffer. */ in = pack_sg_list(chan->sg, out, - VIRTQUEUE_NUM, req->rc->sdata, in_hdr_len); + VIRTQUEUE_NUM, req->rc.sdata, in_hdr_len); if (in) sgs[out_sgs + in_sgs++] = chan->sg + out; diff --git a/net/9p/trans_xen.c b/net/9p/trans_xen.c index c2d54ac76bfd..1a5b38892eb4 100644 --- a/net/9p/trans_xen.c +++ b/net/9p/trans_xen.c @@ -141,7 +141,7 @@ static int p9_xen_request(struct p9_client *client, struct p9_req_t *p9_req) struct xen_9pfs_front_priv *priv = NULL; RING_IDX cons, prod, masked_cons, masked_prod; unsigned long flags; - u32 size = p9_req->tc->size; + u32 size = p9_req->tc.size; struct xen_9pfs_dataring *ring; int num; @@ -154,7 +154,7 @@ static int p9_xen_request(struct p9_client *client, struct p9_req_t *p9_req) if (!priv || priv->client != client) return -EINVAL; - num = p9_req->tc->tag % priv->num_rings; + num = p9_req->tc.tag % priv->num_rings; ring = &priv->rings[num]; again: @@ -176,7 +176,7 @@ static int p9_xen_request(struct p9_client *client, struct p9_req_t *p9_req) masked_prod = xen_9pfs_mask(prod, XEN_9PFS_RING_SIZE); masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE); - xen_9pfs_write_packet(ring->data.out, p9_req->tc->sdata, size, + xen_9pfs_write_packet(ring->data.out, p9_req->tc.sdata, size, &masked_prod, masked_cons, XEN_9PFS_RING_SIZE); p9_req->status = REQ_STATUS_SENT; @@ -229,12 +229,12 @@ static void p9_xen_response(struct work_struct *work) continue; } - memcpy(req->rc, &h, sizeof(h)); - req->rc->offset = 0; + memcpy(&req->rc, &h, sizeof(h)); + req->rc.offset = 0; masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE); /* Then, read the whole packet (including the header) */ - xen_9pfs_read_packet(req->rc->sdata, ring->data.in, h.size, + xen_9pfs_read_packet(req->rc.sdata, ring->data.in, h.size, masked_prod, &masked_cons, XEN_9PFS_RING_SIZE); From patchwork Thu Aug 9 14:33:56 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Dominique Martinet X-Patchwork-Id: 10561475 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 595F013B4 for ; Thu, 9 Aug 2018 14:34:26 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 3DE982B559 for ; Thu, 9 Aug 2018 14:34:26 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 3046C2B55F; Thu, 9 Aug 2018 14:34:26 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 96B512B559 for ; Thu, 9 Aug 2018 14:34:25 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1732236AbeHIQ7W (ORCPT ); Thu, 9 Aug 2018 12:59:22 -0400 Received: from nautica.notk.org ([91.121.71.147]:37673 "EHLO nautica.notk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1731227AbeHIQ7W (ORCPT ); Thu, 9 Aug 2018 12:59:22 -0400 Received: by nautica.notk.org (Postfix, from userid 1001) id 692BBC01A; Thu, 9 Aug 2018 16:34:11 +0200 (CEST) From: Dominique Martinet To: v9fs-developer@lists.sourceforge.net Cc: Dominique Martinet , linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, Matthew Wilcox , Greg Kurz , Jun Piao Subject: [PATCH v3 2/2] net/9p: add a per-client fcall kmem_cache Date: Thu, 9 Aug 2018 16:33:56 +0200 Message-Id: <1533825236-22896-2-git-send-email-asmadeus@codewreck.org> X-Mailer: git-send-email 1.7.10.4 In-Reply-To: <1533825236-22896-1-git-send-email-asmadeus@codewreck.org> References: <1533177452-2165-1-git-send-email-asmadeus@codewreck.org> <1533825236-22896-1-git-send-email-asmadeus@codewreck.org> Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP From: Dominique Martinet Having a specific cache for the fcall allocations helps speed up allocations a bit, especially in case of non-"round" msizes. The caches will automatically be merged if there are multiple caches of items with the same size so we do not need to try to share a cache between different clients of the same size. Since the msize is negotiated with the server, only allocate the cache after that negotiation has happened - previous allocations or allocations of different sizes (e.g. zero-copy fcall) are made with kmalloc directly. Signed-off-by: Dominique Martinet Cc: Matthew Wilcox Cc: Greg Kurz Cc: Jun Piao Acked-by: Jun Piao --- v2: - Add a pointer to the cache in p9_fcall to make sure a buffer allocated with kmalloc gets freed with kfree and vice-versa This could have been smaller with a bool but this spares having to look at the client so looked a bit cleaner, I'm expecting this patch will need a v3 one way or another so I went for the bolder approach - please say if you think a smaller item is better ; I *think* nothing relies on this being ordered the same way as the data on the wire (struct isn't packed anyway) so we can move id after tag and add another u8 to not have any overhead - added likely() to cache existence check in allocation, but nothing for msize check or free because of zc request being of different size v3: - defined the userdata region in the cache, as read or write calls can access the buffer directly (lead to warnings with HARDENED_USERCOPY=y) I didn't get any comment on v2 for this patch, but I'm still not fully happy with this in all honesty. Part of me believes we might be better off always allocating from the cache even for zero-copy headers, it's a waste of space but the buffers are there and being reused constantly for non-zc calls, and mixing kmallocs in could lead to these buffers being really freed and reallocated all the time instead of reusing the slab buffers continuously. That would let me move the likely() for the fast path, with the only exception being the TVERSION initial call on mount, for which I still don't have any nice idea on how to handle, using a different size in p9_client_rpc or prepare_req if the type is TVERSION and I'm really not happy with that... include/net/9p/9p.h | 4 ++++ include/net/9p/client.h | 1 + net/9p/client.c | 40 +++++++++++++++++++++++++++++++++++----- 3 files changed, 40 insertions(+), 5 deletions(-) diff --git a/include/net/9p/9p.h b/include/net/9p/9p.h index e23896116d9a..645266b74652 100644 --- a/include/net/9p/9p.h +++ b/include/net/9p/9p.h @@ -336,6 +336,9 @@ enum p9_qid_t { #define P9_NOFID (u32)(~0) #define P9_MAXWELEM 16 +/* Minimal header size: len + id + tag */ +#define P9_HDRSZ 7 + /* ample room for Twrite/Rread header */ #define P9_IOHDRSZ 24 @@ -558,6 +561,7 @@ struct p9_fcall { size_t offset; size_t capacity; + struct kmem_cache *cache; u8 *sdata; }; diff --git a/include/net/9p/client.h b/include/net/9p/client.h index c2671d40bb6b..735f3979d559 100644 --- a/include/net/9p/client.h +++ b/include/net/9p/client.h @@ -123,6 +123,7 @@ struct p9_client { struct p9_trans_module *trans_mod; enum p9_trans_status status; void *trans; + struct kmem_cache *fcall_cache; union { struct { diff --git a/net/9p/client.c b/net/9p/client.c index ed78751aee7c..6c57ab1294d7 100644 --- a/net/9p/client.c +++ b/net/9p/client.c @@ -192,6 +192,9 @@ static int parse_opts(char *opts, struct p9_client *clnt) goto free_and_return; } + if (clnt->trans_mod) { + pr_warn("Had trans %s\n", clnt->trans_mod->name); + } v9fs_put_trans(clnt->trans_mod); clnt->trans_mod = v9fs_get_trans_by_name(s); if (clnt->trans_mod == NULL) { @@ -231,9 +234,16 @@ static int parse_opts(char *opts, struct p9_client *clnt) return ret; } -static int p9_fcall_init(struct p9_fcall *fc, int alloc_msize) +static int p9_fcall_init(struct p9_client *c, struct p9_fcall *fc, + int alloc_msize) { - fc->sdata = kmalloc(alloc_msize, GFP_NOFS); + if (likely(c->fcall_cache) && alloc_msize == c->msize) { + fc->sdata = kmem_cache_alloc(c->fcall_cache, GFP_NOFS); + fc->cache = c->fcall_cache; + } else { + fc->sdata = kmalloc(alloc_msize, GFP_NOFS); + fc->cache = NULL; + } if (!fc->sdata) return -ENOMEM; fc->capacity = alloc_msize; @@ -242,7 +252,16 @@ static int p9_fcall_init(struct p9_fcall *fc, int alloc_msize) void p9_fcall_fini(struct p9_fcall *fc) { - kfree(fc->sdata); + /* sdata can be NULL for interrupted requests in trans_rdma, + * and kmem_cache_free does not do NULL-check for us + */ + if (unlikely(!fc->sdata)) + return; + + if (fc->cache) + kmem_cache_free(fc->cache, fc->sdata); + else + kfree(fc->sdata); } EXPORT_SYMBOL(p9_fcall_fini); @@ -267,9 +286,9 @@ p9_tag_alloc(struct p9_client *c, int8_t type, unsigned int max_size) if (!req) return NULL; - if (p9_fcall_init(&req->tc, alloc_msize)) + if (p9_fcall_init(c, &req->tc, alloc_msize)) goto free_req; - if (p9_fcall_init(&req->rc, alloc_msize)) + if (p9_fcall_init(c, &req->rc, alloc_msize)) goto free; p9pdu_reset(&req->tc); @@ -951,6 +970,7 @@ struct p9_client *p9_client_create(const char *dev_name, char *options) clnt->trans_mod = NULL; clnt->trans = NULL; + clnt->fcall_cache = NULL; client_id = utsname()->nodename; memcpy(clnt->name, client_id, strlen(client_id) + 1); @@ -987,6 +1007,15 @@ struct p9_client *p9_client_create(const char *dev_name, char *options) if (err) goto close_trans; + /* P9_HDRSZ + 4 is the smallest packet header we can have that is + * followed by data accessed from userspace by read + */ + clnt->fcall_cache = + kmem_cache_create_usercopy("9p-fcall-cache", clnt->msize, + 0, 0, P9_HDRSZ + 4, + clnt->msize - (P9_HDRSZ + 4), + NULL); + return clnt; close_trans: @@ -1018,6 +1047,7 @@ void p9_client_destroy(struct p9_client *clnt) p9_tag_cleanup(clnt); + kmem_cache_destroy(clnt->fcall_cache); kfree(clnt); } EXPORT_SYMBOL(p9_client_destroy);