From: Peng Tao <bergwolf@gmail.com>
To: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-kernel@vger.kernel.org, Peng Tao <bergwolf@gmail.com>,
Andreas Dilger <andreas.dilger@intel.com>
Subject: [PATCH 01/13] staging/lustre: remove server macros from lustre_net.h
Date: Mon, 9 Dec 2013 22:56:53 +0800 [thread overview]
Message-ID: <1386601025-422-2-git-send-email-bergwolf@gmail.com> (raw)
In-Reply-To: <1386601025-422-1-git-send-email-bergwolf@gmail.com>
These macros are only used on server side.
Cc: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <bergwolf@gmail.com>
---
drivers/staging/lustre/lustre/include/lustre_net.h | 234 --------------------
1 file changed, 234 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_net.h b/drivers/staging/lustre/lustre/include/lustre_net.h
index 6c5479b..d8d0880 100644
--- a/drivers/staging/lustre/lustre/include/lustre_net.h
+++ b/drivers/staging/lustre/lustre/include/lustre_net.h
@@ -264,241 +264,7 @@
#define LDLM_MAXREQSIZE (5 * 1024)
#define LDLM_MAXREPSIZE (1024)
- /*
- * MDS threads constants:
- *
- * Please see examples in "Thread Constants", MDS threads number will be at
- * the comparable level of old versions, unless the server has many cores.
- */
-#ifndef MDS_MAX_THREADS
-#define MDS_MAX_THREADS 1024
-#define MDS_MAX_OTHR_THREADS 256
-
-#else /* MDS_MAX_THREADS */
-#if MDS_MAX_THREADS < PTLRPC_NTHRS_INIT
-#undef MDS_MAX_THREADS
-#define MDS_MAX_THREADS PTLRPC_NTHRS_INIT
-#endif
-#define MDS_MAX_OTHR_THREADS max(PTLRPC_NTHRS_INIT, MDS_MAX_THREADS / 2)
-#endif
-
-/* default service */
-#define MDS_THR_FACTOR 8
-#define MDS_NTHRS_INIT PTLRPC_NTHRS_INIT
-#define MDS_NTHRS_MAX MDS_MAX_THREADS
-#define MDS_NTHRS_BASE min(64, MDS_NTHRS_MAX)
-
-/* read-page service */
-#define MDS_RDPG_THR_FACTOR 4
-#define MDS_RDPG_NTHRS_INIT PTLRPC_NTHRS_INIT
-#define MDS_RDPG_NTHRS_MAX MDS_MAX_OTHR_THREADS
-#define MDS_RDPG_NTHRS_BASE min(48, MDS_RDPG_NTHRS_MAX)
-
-/* these should be removed when we remove setattr service in the future */
-#define MDS_SETA_THR_FACTOR 4
-#define MDS_SETA_NTHRS_INIT PTLRPC_NTHRS_INIT
-#define MDS_SETA_NTHRS_MAX MDS_MAX_OTHR_THREADS
-#define MDS_SETA_NTHRS_BASE min(48, MDS_SETA_NTHRS_MAX)
-
-/* non-affinity threads */
-#define MDS_OTHR_NTHRS_INIT PTLRPC_NTHRS_INIT
-#define MDS_OTHR_NTHRS_MAX MDS_MAX_OTHR_THREADS
-
-#define MDS_NBUFS 64
-
-/**
- * Assume file name length = FNAME_MAX = 256 (true for ext3).
- * path name length = PATH_MAX = 4096
- * LOV MD size max = EA_MAX = 24 * 2000
- * (NB: 24 is size of lov_ost_data)
- * LOV LOGCOOKIE size max = 32 * 2000
- * (NB: 32 is size of llog_cookie)
- * symlink: FNAME_MAX + PATH_MAX <- largest
- * link: FNAME_MAX + PATH_MAX (mds_rec_link < mds_rec_create)
- * rename: FNAME_MAX + FNAME_MAX
- * open: FNAME_MAX + EA_MAX
- *
- * MDS_MAXREQSIZE ~= 4736 bytes =
- * lustre_msg + ldlm_request + mdt_body + mds_rec_create + FNAME_MAX + PATH_MAX
- * MDS_MAXREPSIZE ~= 8300 bytes = lustre_msg + llog_header
- *
- * Realistic size is about 512 bytes (20 character name + 128 char symlink),
- * except in the open case where there are a large number of OSTs in a LOV.
- */
-#define MDS_MAXREQSIZE (5 * 1024) /* >= 4736 */
-#define MDS_MAXREPSIZE (9 * 1024) /* >= 8300 */
-
-/**
- * MDS incoming request with LOV EA
- * 24 = sizeof(struct lov_ost_data), i.e: replay of opencreate
- */
-#define MDS_LOV_MAXREQSIZE max(MDS_MAXREQSIZE, \
- 362 + LOV_MAX_STRIPE_COUNT * 24)
-/**
- * MDS outgoing reply with LOV EA
- *
- * NB: max reply size Lustre 2.4+ client can get from old MDS is:
- * LOV_MAX_STRIPE_COUNT * (llog_cookie + lov_ost_data) + extra bytes
- *
- * but 2.4 or later MDS will never send reply with llog_cookie to any
- * version client. This macro is defined for server side reply buffer size.
- */
-#define MDS_LOV_MAXREPSIZE MDS_LOV_MAXREQSIZE
-
-/**
- * This is the size of a maximum REINT_SETXATTR request:
- *
- * lustre_msg 56 (32 + 4 x 5 + 4)
- * ptlrpc_body 184
- * mdt_rec_setxattr 136
- * lustre_capa 120
- * name 256 (XATTR_NAME_MAX)
- * value 65536 (XATTR_SIZE_MAX)
- */
-#define MDS_EA_MAXREQSIZE 66288
-
-/**
- * These are the maximum request and reply sizes (rounded up to 1 KB
- * boundaries) for the "regular" MDS_REQUEST_PORTAL and MDS_REPLY_PORTAL.
- */
-#define MDS_REG_MAXREQSIZE (((max(MDS_EA_MAXREQSIZE, \
- MDS_LOV_MAXREQSIZE) + 1023) >> 10) << 10)
-#define MDS_REG_MAXREPSIZE MDS_REG_MAXREQSIZE
-
-/**
- * The update request includes all of updates from the create, which might
- * include linkea (4K maxim), together with other updates, we set it to 9K:
- * lustre_msg + ptlrpc_body + UPDATE_BUF_SIZE (8K)
- */
-#define MDS_OUT_MAXREQSIZE (9 * 1024)
-#define MDS_OUT_MAXREPSIZE MDS_MAXREPSIZE
-
-/** MDS_BUFSIZE = max_reqsize (w/o LOV EA) + max sptlrpc payload size */
-#define MDS_BUFSIZE max(MDS_MAXREQSIZE + SPTLRPC_MAX_PAYLOAD, \
- 8 * 1024)
-
-/**
- * MDS_REG_BUFSIZE should at least be MDS_REG_MAXREQSIZE + SPTLRPC_MAX_PAYLOAD.
- * However, we need to allocate a much larger buffer for it because LNet
- * requires each MD(rqbd) has at least MDS_REQ_MAXREQSIZE bytes left to avoid
- * dropping of maximum-sized incoming request. So if MDS_REG_BUFSIZE is only a
- * little larger than MDS_REG_MAXREQSIZE, then it can only fit in one request
- * even there are about MDS_REG_MAX_REQSIZE bytes left in a rqbd, and memory
- * utilization is very low.
- *
- * In the meanwhile, size of rqbd can't be too large, because rqbd can't be
- * reused until all requests fit in it have been processed and released,
- * which means one long blocked request can prevent the rqbd be reused.
- * Now we set request buffer size to 160 KB, so even each rqbd is unlinked
- * from LNet with unused 65 KB, buffer utilization will be about 59%.
- * Please check LU-2432 for details.
- */
-#define MDS_REG_BUFSIZE max(MDS_REG_MAXREQSIZE + SPTLRPC_MAX_PAYLOAD, \
- 160 * 1024)
-
-/**
- * MDS_OUT_BUFSIZE = max_out_reqsize + max sptlrpc payload (~1K) which is
- * about 10K, for the same reason as MDS_REG_BUFSIZE, we also give some
- * extra bytes to each request buffer to improve buffer utilization rate.
- */
-#define MDS_OUT_BUFSIZE max(MDS_OUT_MAXREQSIZE + SPTLRPC_MAX_PAYLOAD, \
- 24 * 1024)
-
-/** FLD_MAXREQSIZE == lustre_msg + __u32 padding + ptlrpc_body + opc */
-#define FLD_MAXREQSIZE (160)
-
-/** FLD_MAXREPSIZE == lustre_msg + ptlrpc_body */
-#define FLD_MAXREPSIZE (152)
-#define FLD_BUFSIZE (1 << 12)
-
-/**
- * SEQ_MAXREQSIZE == lustre_msg + __u32 padding + ptlrpc_body + opc + lu_range +
- * __u32 padding */
-#define SEQ_MAXREQSIZE (160)
-
-/** SEQ_MAXREPSIZE == lustre_msg + ptlrpc_body + lu_range */
-#define SEQ_MAXREPSIZE (152)
-#define SEQ_BUFSIZE (1 << 12)
-
-/** MGS threads must be >= 3, see bug 22458 comment #28 */
-#define MGS_NTHRS_INIT (PTLRPC_NTHRS_INIT + 1)
-#define MGS_NTHRS_MAX 32
-
-#define MGS_NBUFS 64
-#define MGS_BUFSIZE (8 * 1024)
-#define MGS_MAXREQSIZE (7 * 1024)
-#define MGS_MAXREPSIZE (9 * 1024)
-
- /*
- * OSS threads constants:
- *
- * Given 8 as factor and 64 as base threads number
- *
- * example 1):
- * On 8-core server configured to 2 partitions, we will have
- * 64 + 8 * 4 = 96 threads for each partition, 192 total threads.
- *
- * example 2):
- * On 32-core machine configured to 4 partitions, we will have
- * 64 + 8 * 8 = 112 threads for each partition, so total threads number
- * will be 112 * 4 = 448.
- *
- * example 3):
- * On 64-core machine configured to 4 partitions, we will have
- * 64 + 16 * 8 = 192 threads for each partition, so total threads number
- * will be 192 * 4 = 768 which is above limit OSS_NTHRS_MAX(512), so we
- * cut off the value to OSS_NTHRS_MAX(512) / 4 which is 128 threads
- * for each partition.
- *
- * So we can see that with these constants, threads number wil be at the
- * similar level of old versions, unless the server has many cores.
- */
- /* depress threads factor for VM with small memory size */
-#define OSS_THR_FACTOR min_t(int, 8, \
- NUM_CACHEPAGES >> (28 - PAGE_CACHE_SHIFT))
-#define OSS_NTHRS_INIT (PTLRPC_NTHRS_INIT + 1)
-#define OSS_NTHRS_BASE 64
-#define OSS_NTHRS_MAX 512
-
-/* threads for handling "create" request */
-#define OSS_CR_THR_FACTOR 1
-#define OSS_CR_NTHRS_INIT PTLRPC_NTHRS_INIT
-#define OSS_CR_NTHRS_BASE 8
-#define OSS_CR_NTHRS_MAX 64
-
-/**
- * OST_IO_MAXREQSIZE ~=
- * lustre_msg + ptlrpc_body + obdo + obd_ioobj +
- * DT_MAX_BRW_PAGES * niobuf_remote
- *
- * - single object with 16 pages is 512 bytes
- * - OST_IO_MAXREQSIZE must be at least 1 page of cookies plus some spillover
- * - Must be a multiple of 1024
- * - actual size is about 18K
- */
-#define _OST_MAXREQSIZE_SUM (sizeof(struct lustre_msg) + \
- sizeof(struct ptlrpc_body) + \
- sizeof(struct obdo) + \
- sizeof(struct obd_ioobj) + \
- sizeof(struct niobuf_remote) * DT_MAX_BRW_PAGES)
-/**
- * FIEMAP request can be 4K+ for now
- */
#define OST_MAXREQSIZE (5 * 1024)
-#define OST_IO_MAXREQSIZE max_t(int, OST_MAXREQSIZE, \
- (((_OST_MAXREQSIZE_SUM - 1) | (1024 - 1)) + 1))
-
-#define OST_MAXREPSIZE (9 * 1024)
-#define OST_IO_MAXREPSIZE OST_MAXREPSIZE
-
-#define OST_NBUFS 64
-/** OST_BUFSIZE = max_reqsize + max sptlrpc payload size */
-#define OST_BUFSIZE max_t(int, OST_MAXREQSIZE + 1024, 16 * 1024)
-/**
- * OST_IO_MAXREQSIZE is 18K, giving extra 46K can increase buffer utilization
- * rate of request buffer, please check comment of MDS_LOV_BUFSIZE for details.
- */
-#define OST_IO_BUFSIZE max_t(int, OST_IO_MAXREQSIZE + 1024, 64 * 1024)
/* Macro to hide a typecast. */
#define ptlrpc_req_async_args(req) ((void *)&req->rq_async_args)
--
1.7.9.5
next prev parent reply other threads:[~2013-12-09 15:00 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-12-09 14:56 [PATCH 00/13] staging/lustre: sync with external tree, set 3 Peng Tao
2013-12-09 14:56 ` Peng Tao [this message]
2013-12-09 14:56 ` [PATCH 02/13] staging/lustre/lnet: Add LNet Router Priority parameter Peng Tao
2013-12-09 14:56 ` [PATCH 03/13] staging/lustre/api: HSM import uses new released pattern Peng Tao
2013-12-09 14:56 ` [PATCH 04/13] staging/lustre/target: move OUT to the unified target code Peng Tao
2013-12-09 14:56 ` [PATCH 05/13] staging/lustre/seq: remove seq_{query,handle} Peng Tao
2013-12-09 14:56 ` [PATCH 06/13] staging/lustre/llite: remove ll_d_root_ops Peng Tao
2013-12-11 2:32 ` Greg Kroah-Hartman
2013-12-11 3:23 ` Peng Tao
2013-12-09 14:56 ` [PATCH 07/13] staging/lustre/llite: don't d_add for create only files Peng Tao
2013-12-09 14:57 ` [PATCH 08/13] staging/lustre/llite: pass correct pointer to obd_iocontrol() Peng Tao
2013-12-09 14:57 ` [PATCH 09/13] staging/lustre/idl: remove LASSERT/CLASSERT from lustre_idl.h Peng Tao
2013-12-09 14:57 ` [PATCH 10/13] staging/lustre/mgs: set_param -P option that sets value permanently Peng Tao
2013-12-09 14:57 ` [PATCH 11/13] staging/lustre/obdclass: remove extra break in class_process_config Peng Tao
2013-12-09 14:57 ` [PATCH 12/13] staging/lustre/lustre_user.h: remove obsolete comments Peng Tao
2013-12-09 14:57 ` [PATCH 13/13] staging/lustre/ptlrpc: flock deadlock detection does not work Peng Tao
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1386601025-422-2-git-send-email-bergwolf@gmail.com \
--to=bergwolf@gmail.com \
--cc=andreas.dilger@intel.com \
--cc=gregkh@linuxfoundation.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.