From: kbuild test robot <lkp@intel.com>
To: Serge Semin <fancer.lancer@gmail.com>
Cc: kbuild-all@01.org, jdmason@kudzu.us, dave.jiang@intel.com,
Allen.Hubbe@emc.com, Shyam-sundar.S-k@amd.com,
Xiangliang.Yu@amd.com, Sergey.Semin@t-platforms.ru,
linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org,
Serge Semin <fancer.lancer@gmail.com>
Subject: Re: [PATCH v2] NTB: ntb_perf: Add full multi-port NTB API support
Date: Mon, 4 Dec 2017 01:55:58 +0800 [thread overview]
Message-ID: <201712040124.tuJQTbll%fengguang.wu@intel.com> (raw)
In-Reply-To: <20171130213809.20984-1-fancer.lancer@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 12518 bytes --]
Hi Serge,
I love your patch! Yet something to improve:
[auto build test ERROR on v4.15-rc1]
[cannot apply to ntb/ntb-next next-20171201]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Serge-Semin/NTB-ntb_perf-Add-full-multi-port-NTB-API-support/20171204-011126
config: x86_64-randconfig-x017-201749 (attached as .config)
compiler: gcc-7 (Debian 7.2.0-12) 7.2.1 20171025
reproduce:
# save the attached .config to linux build tree
make ARCH=x86_64
All error/warnings (new ones prefixed by >>):
drivers/ntb//test/ntb_perf.c: In function 'perf_msg_cmd_send':
>> drivers/ntb//test/ntb_perf.c:375:3: error: implicit declaration of function 'ntb_peer_msg_write'; did you mean 'ntb_peer_spad_write'? [-Werror=implicit-function-declaration]
ntb_peer_msg_write(perf->ntb, peer->pidx, PERF_MSG_LDATA,
^~~~~~~~~~~~~~~~~~
ntb_peer_spad_write
drivers/ntb//test/ntb_perf.c: In function 'perf_msg_cmd_recv':
>> drivers/ntb//test/ntb_perf.c:408:32: warning: passing argument 2 of 'ntb_msg_read' makes integer from pointer without a cast [-Wint-conversion]
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_CMD);
^~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: expected 'int' but argument is of type 'int *'
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
>> drivers/ntb//test/ntb_perf.c:408:8: error: too few arguments to function 'ntb_msg_read'
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_CMD);
^~~~~~~~~~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: declared here
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
drivers/ntb//test/ntb_perf.c:411:32: warning: passing argument 2 of 'ntb_msg_read' makes integer from pointer without a cast [-Wint-conversion]
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_LDATA);
^~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: expected 'int' but argument is of type 'int *'
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
>> drivers/ntb//test/ntb_perf.c:229:25: warning: passing argument 3 of 'ntb_msg_read' makes pointer from integer without a cast [-Wint-conversion]
#define PERF_MSG_LDATA 1
^
>> drivers/ntb//test/ntb_perf.c:411:38: note: in expansion of macro 'PERF_MSG_LDATA'
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_LDATA);
^~~~~~~~~~~~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: expected 'int *' but argument is of type 'int'
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
drivers/ntb//test/ntb_perf.c:411:8: error: too few arguments to function 'ntb_msg_read'
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_LDATA);
^~~~~~~~~~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: declared here
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
drivers/ntb//test/ntb_perf.c:414:32: warning: passing argument 2 of 'ntb_msg_read' makes integer from pointer without a cast [-Wint-conversion]
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_HDATA);
^~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: expected 'int' but argument is of type 'int *'
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
drivers/ntb//test/ntb_perf.c:230:25: warning: passing argument 3 of 'ntb_msg_read' makes pointer from integer without a cast [-Wint-conversion]
#define PERF_MSG_HDATA 2
^
>> drivers/ntb//test/ntb_perf.c:414:38: note: in expansion of macro 'PERF_MSG_HDATA'
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_HDATA);
^~~~~~~~~~~~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: expected 'int *' but argument is of type 'int'
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
drivers/ntb//test/ntb_perf.c:414:8: error: too few arguments to function 'ntb_msg_read'
val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_HDATA);
^~~~~~~~~~~~
In file included from drivers/ntb//test/ntb_perf.c:89:0:
include/linux/ntb.h:1473:19: note: declared here
static inline int ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx,
^~~~~~~~~~~~
cc1: some warnings being treated as errors
vim +375 drivers/ntb//test/ntb_perf.c
209
210 /*
211 * Scratchpads-base commands interface
212 */
213 #define PERF_SPAD_CNT(_pcnt) \
214 (3*((_pcnt) + 1))
215 #define PERF_SPAD_CMD(_gidx) \
216 (3*(_gidx))
217 #define PERF_SPAD_LDATA(_gidx) \
218 (3*(_gidx) + 1)
219 #define PERF_SPAD_HDATA(_gidx) \
220 (3*(_gidx) + 2)
221 #define PERF_SPAD_NOTIFY(_gidx) \
222 (BIT_ULL(_gidx))
223
224 /*
225 * Messages-base commands interface
226 */
227 #define PERF_MSG_CNT 3
228 #define PERF_MSG_CMD 0
> 229 #define PERF_MSG_LDATA 1
230 #define PERF_MSG_HDATA 2
231
232 /*==============================================================================
233 * Static data declarations
234 *==============================================================================
235 */
236
237 static struct dentry *perf_dbgfs_topdir;
238
239 static struct workqueue_struct *perf_wq __read_mostly;
240
241 /*==============================================================================
242 * NTB cross-link commands execution service
243 *==============================================================================
244 */
245
246 static void perf_terminate_test(struct perf_ctx *perf);
247
248 static inline bool perf_link_is_up(struct perf_peer *peer)
249 {
250 u64 link;
251
252 link = ntb_link_is_up(peer->perf->ntb, NULL, NULL);
253 return !!(link & BIT_ULL_MASK(peer->pidx));
254 }
255
256 static int perf_spad_cmd_send(struct perf_peer *peer, enum perf_cmd cmd,
257 u64 data)
258 {
259 struct perf_ctx *perf = peer->perf;
260 int try;
261 u32 sts;
262
263 dev_dbg(&perf->ntb->dev, "CMD send: %d 0x%llx\n", cmd, data);
264
265 /*
266 * Perform predefined number of attempts before give up.
267 * We are sending the data to the port specific scratchpad, so
268 * to prevent a multi-port access race-condition. Additionally
269 * there is no need in local locking since only thread-safe
270 * service work is using this method.
271 */
272 for (try = 0; try < MSG_TRIES; try++) {
273 if (!perf_link_is_up(peer))
274 return -ENOLINK;
275
276 sts = ntb_peer_spad_read(perf->ntb, peer->pidx,
277 PERF_SPAD_CMD(perf->gidx));
278 if (le32_to_cpu(sts) != PERF_CMD_INVAL) {
279 usleep_range(MSG_UDELAY_LOW, MSG_UDELAY_HIGH);
280 continue;
281 }
282
283 ntb_peer_spad_write(perf->ntb, peer->pidx,
284 PERF_SPAD_LDATA(perf->gidx),
285 cpu_to_le32(lower_32_bits(data)));
286 ntb_peer_spad_write(perf->ntb, peer->pidx,
287 PERF_SPAD_HDATA(perf->gidx),
288 cpu_to_le32(upper_32_bits(data)));
289 mmiowb();
290 ntb_peer_spad_write(perf->ntb, peer->pidx,
291 PERF_SPAD_CMD(perf->gidx),
292 cpu_to_le32(cmd));
293 mmiowb();
294 ntb_peer_db_set(perf->ntb, PERF_SPAD_NOTIFY(peer->gidx));
295
296 dev_dbg(&perf->ntb->dev, "DB ring peer %#llx\n",
297 PERF_SPAD_NOTIFY(peer->gidx));
298
299 break;
300 }
301
302 return try < MSG_TRIES ? 0 : -EAGAIN;
303 }
304
305 static int perf_spad_cmd_recv(struct perf_ctx *perf, int *pidx,
306 enum perf_cmd *cmd, u64 *data)
307 {
308 struct perf_peer *peer;
309 u32 val;
310
311 ntb_db_clear(perf->ntb, PERF_SPAD_NOTIFY(perf->gidx));
312
313 /*
314 * We start scanning all over, since cleared DB may have been set
315 * by any peer. Yes, it makes peer with smaller index being
316 * serviced with greater priority, but it's convenient for spad
317 * and message code unification and simplicity.
318 */
319 for (*pidx = 0; *pidx < perf->pcnt; (*pidx)++) {
320 peer = &perf->peers[*pidx];
321
322 if (!perf_link_is_up(peer))
323 continue;
324
325 val = ntb_spad_read(perf->ntb, PERF_SPAD_CMD(peer->gidx));
326 val = le32_to_cpu(val);
327 if (val == PERF_CMD_INVAL)
328 continue;
329
330 *cmd = val;
331
332 val = ntb_spad_read(perf->ntb, PERF_SPAD_LDATA(peer->gidx));
333 *data = le32_to_cpu(val);
334
335 val = ntb_spad_read(perf->ntb, PERF_SPAD_HDATA(peer->gidx));
336 *data |= (u64)le32_to_cpu(val) << 32;
337
338 /* Next command can be retrieved from now */
339 ntb_spad_write(perf->ntb, PERF_SPAD_CMD(peer->gidx),
340 cpu_to_le32(PERF_CMD_INVAL));
341
342 dev_dbg(&perf->ntb->dev, "CMD recv: %d 0x%llx\n", *cmd, *data);
343
344 return 0;
345 }
346
347 return -ENODATA;
348 }
349
350 static int perf_msg_cmd_send(struct perf_peer *peer, enum perf_cmd cmd,
351 u64 data)
352 {
353 struct perf_ctx *perf = peer->perf;
354 int try, ret;
355 u64 outbits;
356
357 dev_dbg(&perf->ntb->dev, "CMD send: %d 0x%llx\n", cmd, data);
358
359 /*
360 * Perform predefined number of attempts before give up. Message
361 * registers are free of race-condition problem when accessed
362 * from different ports, so we don't need splitting registers
363 * by global device index. We also won't have local locking,
364 * since the method is used from service work only.
365 */
366 outbits = ntb_msg_outbits(perf->ntb);
367 for (try = 0; try < MSG_TRIES; try++) {
368 if (!perf_link_is_up(peer))
369 return -ENOLINK;
370
371 ret = ntb_msg_clear_sts(perf->ntb, outbits);
372 if (ret)
373 return ret;
374
> 375 ntb_peer_msg_write(perf->ntb, peer->pidx, PERF_MSG_LDATA,
376 cpu_to_le32(lower_32_bits(data)));
377
378 if (ntb_msg_read_sts(perf->ntb) & outbits) {
379 usleep_range(MSG_UDELAY_LOW, MSG_UDELAY_HIGH);
380 continue;
381 }
382
383 ntb_peer_msg_write(perf->ntb, peer->pidx, PERF_MSG_HDATA,
384 cpu_to_le32(upper_32_bits(data)));
385 mmiowb();
386
387 /* This call shall trigger peer message event */
388 ntb_peer_msg_write(perf->ntb, peer->pidx, PERF_MSG_CMD,
389 cpu_to_le32(cmd));
390
391 break;
392 }
393
394 return try < MSG_TRIES ? 0 : -EAGAIN;
395 }
396
397 static int perf_msg_cmd_recv(struct perf_ctx *perf, int *pidx,
398 enum perf_cmd *cmd, u64 *data)
399 {
400 u64 inbits;
401 u32 val;
402
403 inbits = ntb_msg_inbits(perf->ntb);
404
405 if (hweight64(ntb_msg_read_sts(perf->ntb) & inbits) < 3)
406 return -ENODATA;
407
> 408 val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_CMD);
409 *cmd = le32_to_cpu(val);
410
> 411 val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_LDATA);
412 *data = le32_to_cpu(val);
413
> 414 val = ntb_msg_read(perf->ntb, pidx, PERF_MSG_HDATA);
415 *data |= (u64)le32_to_cpu(val) << 32;
416
417 /* Next command can be retrieved from now */
418 ntb_msg_clear_sts(perf->ntb, inbits);
419
420 dev_dbg(&perf->ntb->dev, "CMD recv: %d 0x%llx\n", *cmd, *data);
421
422 return 0;
423 }
424
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 21126 bytes --]
next prev parent reply other threads:[~2017-12-03 17:56 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-08 10:16 [PATCH] NTB: ntb_perf: Add full multi-port NTB API support Serge Semin
2017-11-30 21:38 ` [PATCH v2] " Serge Semin
2017-12-03 17:55 ` kbuild test robot [this message]
2017-12-03 17:58 ` kbuild test robot
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=201712040124.tuJQTbll%fengguang.wu@intel.com \
--to=lkp@intel.com \
--cc=Allen.Hubbe@emc.com \
--cc=Sergey.Semin@t-platforms.ru \
--cc=Shyam-sundar.S-k@amd.com \
--cc=Xiangliang.Yu@amd.com \
--cc=dave.jiang@intel.com \
--cc=fancer.lancer@gmail.com \
--cc=jdmason@kudzu.us \
--cc=kbuild-all@01.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-ntb@googlegroups.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox