On Tue, Apr 14, 2020 at 5:56 PM 'Guenter Roeck' via Clang Built Linux
<clang-built-linux@xxxxxxxxxxxxxxxx> wrote:
>
> On Tue, Apr 14, 2020 at 5:26 PM Nathan Chancellor
> <natechancellor@xxxxxxxxx> wrote:
> >
> > Hi all,
> >
> > Sorry for yet another convergeance on this commit... :/ hopefully this
> > does not continue for much longer. None of the warnings are obviously
> > caused by the patch below.
>> Fixed by commit 85dc2c65e6c9 ("misc: echo: Remove unnecessary
>> parentheses and simplify check for zero").
> >
> No worries.
>
> I noticed that the problems are pretty much all fixed in the upstream
> kernel. I just wasn't sure if it would be worthwhile sending a request
> to stable@ to have them applied to 4.19.y (and if necessary 5.4.y).
> Any suggestions ?
We should strive to be warning free on stable. Thanks for identifying
the fix Nathan.
Greg, Sasha,
Would you please cherry pick 85dc2c65e6c9 to 4.19.y, 4.14.y, 4.9.y,
and 4.4.y (maybe 3.18, didn't check that one)? It applies cleanly and
is a trivial fix for a warning that landed in v4.20-rc1.
`git describe --contains "$tag" | sed 's/~.*//'` is my trick for
finding the first tag that contained a commit.
>
> Thanks,
> Guenter
>
>
> Guenter
>
> > On Wed, Apr 15, 2020 at 06:09:47AM +0800, kbuild test robot wrote:
> > > tree: https://chromium.googlesource.com/chromiumos/third_party/kernel chromeos-4.19
> > > head: e747fc8dfce13c501be9b7032a5a3d8249bc27ab
> > > commit: 12bd892cc0bd4a700dd41aa4b0d1e0cfc128e3e4 [21350/21402] CHROMIUM: Merge 'v4.19.114' into chromeos-4.19
> > > config: x86_64-randconfig-c002-20200414 (attached as .config)
> > > compiler: clang version 11.0.0 (https://github.com/llvm/llvm-project f3da6b7ab5698bb0b9bd208972115dbcbbc59f27)
> > > reproduce:
> > > wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
> > > chmod +x ~/bin/make.cross
> > > git checkout 12bd892cc0bd4a700dd41aa4b0d1e0cfc128e3e4
> > > # save the attached .config to linux build tree
> > > COMPILER=clang make.cross ARCH=x86_64
> > >
> > > If you fix the issue, kindly add following tag as appropriate
> > > Reported-by: kbuild test robot <lkp@xxxxxxxxx>
> > >
> > > All errors (new ones prefixed by >>):
> > >
> > > >> drivers/misc/echo/echo.c:384:27: error: equality comparison with extraneous parentheses [-Werror,-Wparentheses-equality]
> > > if ((ec->nonupdate_dwell == 0)) {
> > > ~~~~~~~~~~~~~~~~~~~~^~~~
> > > drivers/misc/echo/echo.c:384:27: note: remove extraneous parentheses around the comparison to silence this warning
> > > if ((ec->nonupdate_dwell == 0)) {
> > > ~ ^ ~
> > > drivers/misc/echo/echo.c:384:27: note: use '=' to turn this equality comparison into an assignment
> > > if ((ec->nonupdate_dwell == 0)) {
> > > ^~
> > > =
> > > 1 error generated.
> >
> > Fixed by commit 85dc2c65e6c9 ("misc: echo: Remove unnecessary
> > parentheses and simplify check for zero").
> >
> > > >> net/decnet/dn_dev.c:1366:10: error: address of array 'dev->name' will always evaluate to 'true' [-Werror,-Wpointer-bool-conversion]
> > > dev->name ? dev->name : "???",
> > > ~~~~~^~~~ ~
> > > 1 error generated.
> >
> > Fixed by commit 5b9b0a80aa1a ("decnet: Remove unnecessary check for
> > dev->name").
> >
> > > >> drivers/regulator/da903x.c:395:2: error: division by zero is undefined [-Werror,-Wdivision-by-zero]
> > > DA9030_LDO(13, 2100, 2100, 0, INVAL, 0, 0, RCTL11, 3), /* fixed @2.1V */
> > > ^ ~
> > > drivers/regulator/da903x.c:359:2: note: expanded from macro 'DA9030_LDO'
> > > DA903x_LDO(DA9030, _id, min, max, step, vreg, shift, nbits, ereg, ebit)
> > > ^ ~~~~
> > > drivers/regulator/da903x.c:320:39: note: expanded from macro 'DA903x_LDO'
> > > .n_voltages = (step) ? ((max - min) / step + 1) : 1, \
> > > ^ ~~~~
> > > drivers/regulator/da903x.c:415:2: error: division by zero is undefined [-Werror,-Wdivision-by-zero]
> > > DA9034_LDO(5, 3100, 3100, 0, INVAL, 0, 0, OVER3, 7), /* fixed @3.1V */
> > > ^ ~
> > > drivers/regulator/da903x.c:356:2: note: expanded from macro 'DA9034_LDO'
> > > DA903x_LDO(DA9034, _id, min, max, step, vreg, shift, nbits, ereg, ebit)
> > > ^ ~~~~
> > > drivers/regulator/da903x.c:320:39: note: expanded from macro 'DA903x_LDO'
> > > .n_voltages = (step) ? ((max - min) / step + 1) : 1, \
> > > ^ ~~~~
> > > 2 errors generated.
> >
> > "Fixed" by commit 0077aaaeeb69 ("regulator: da903x: don't build with
> > clang"). I do not think the Clang fix ever got finalized.
> >
> > > In file included from drivers/scsi/osd/osd_uld.c:60:
> > > In file included from include/scsi/osd_initiator.h:18:
> > > >> include/scsi/osd_types.h:31:21: error: 'weak' attribute only applies to variables, functions, and classes [-Werror,-Wignored-attributes]
> > > static const struct __weak osd_obj_id osd_root_object = {0, 0};
> > > ^
> > > include/linux/compiler_types.h:219:33: note: expanded from macro '__weak'
> > > #define __weak __attribute__((weak))
> > > ^
> > > 1 error generated.
> >
> > "Fixed" by just deleting the driver in commit 19fcae3d4f2d ("scsi:
> > remove the SCSI OSD library"). I did send a patch to fix it which
> > spurred the deletion, if we wanted to take it in stable if we care:
> >
> > https://lore.kernel.org/lkml/20190126064740.12949-1-natechancellor@xxxxxxxxx/
> >
> > > vim +384 drivers/misc/echo/echo.c
> > >
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 273
> > > 9d8f2d5dfbcc9d drivers/staging/echo/echo.c Tzafrir Cohen 2008-10-12 274 int16_t oslec_update(struct oslec_state *ec, int16_t tx, int16_t rx)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 275 {
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 276 int32_t echo_value;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 277 int clean_bg;
> > > 3ec50be588d773 drivers/staging/echo/echo.c Jesper Juhl 2012-06-27 278 int tmp;
> > > 3ec50be588d773 drivers/staging/echo/echo.c Jesper Juhl 2012-06-27 279 int tmp1;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 280
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 281 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 282 * Input scaling was found be required to prevent problems when tx
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 283 * starts clipping. Another possible way to handle this would be the
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 284 * filter coefficent scaling.
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 285 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 286
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 287 ec->tx = tx;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 288 ec->rx = rx;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 289 tx >>= 1;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 290 rx >>= 1;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 291
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 292 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 293 * Filter DC, 3dB point is 160Hz (I think), note 32 bit precision
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 294 * required otherwise values do not track down to 0. Zero at DC, Pole
> > > 196e76e86a3a90 drivers/staging/echo/echo.c David Rowe 2009-08-23 295 * at (1-Beta) on real axis. Some chip sets (like Si labs) don't
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 296 * need this, but something like a $10 X100P card does. Any DC really
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 297 * slows down convergence.
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 298 *
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 299 * Note: removes some low frequency from the signal, this reduces the
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 300 * speech quality when listening to samples through headphones but may
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 301 * not be obvious through a telephone handset.
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 302 *
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 303 * Note that the 3dB frequency in radians is approx Beta, e.g. for Beta
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 304 * = 2^(-3) = 0.125, 3dB freq is 0.125 rads = 159Hz.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 305 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 306
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 307 if (ec->adaption_mode & ECHO_CAN_USE_RX_HPF) {
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 308 tmp = rx << 15;
> > > 196e76e86a3a90 drivers/staging/echo/echo.c David Rowe 2009-08-23 309
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 310 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 311 * Make sure the gain of the HPF is 1.0. This can still
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 312 * saturate a little under impulse conditions, and it might
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 313 * roll to 32768 and need clipping on sustained peak level
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 314 * signals. However, the scale of such clipping is small, and
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 315 * the error due to any saturation should not markedly affect
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 316 * the downstream processing.
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 317 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 318 tmp -= (tmp >> 4);
> > > 196e76e86a3a90 drivers/staging/echo/echo.c David Rowe 2009-08-23 319
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 320 ec->rx_1 += -(ec->rx_1 >> DC_LOG2BETA) + tmp - ec->rx_2;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 321
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 322 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 323 * hard limit filter to prevent clipping. Note that at this
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 324 * stage rx should be limited to +/- 16383 due to right shift
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 325 * above
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 326 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 327 tmp1 = ec->rx_1 >> 15;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 328 if (tmp1 > 16383)
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 329 tmp1 = 16383;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 330 if (tmp1 < -16383)
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 331 tmp1 = -16383;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 332 rx = tmp1;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 333 ec->rx_2 = tmp;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 334 }
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 335
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 336 /* Block average of power in the filter states. Used for
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 337 adaption power calculation. */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 338
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 339 {
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 340 int new, old;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 341
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 342 /* efficient "out with the old and in with the new" algorithm so
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 343 we don't have to recalculate over the whole block of
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 344 samples. */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 345 new = (int)tx * (int)tx;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 346 old = (int)ec->fir_state.history[ec->fir_state.curr_pos] *
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 347 (int)ec->fir_state.history[ec->fir_state.curr_pos];
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 348 ec->pstates +=
> > > 0f51010e87636e drivers/staging/echo/echo.c David Rowe 2009-05-20 349 ((new - old) + (1 << (ec->log2taps - 1))) >> ec->log2taps;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 350 if (ec->pstates < 0)
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 351 ec->pstates = 0;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 352 }
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 353
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 354 /* Calculate short term average levels using simple single pole IIRs */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 355
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 356 ec->ltxacc += abs(tx) - ec->ltx;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 357 ec->ltx = (ec->ltxacc + (1 << 4)) >> 5;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 358 ec->lrxacc += abs(rx) - ec->lrx;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 359 ec->lrx = (ec->lrxacc + (1 << 4)) >> 5;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 360
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 361 /* Foreground filter */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 362
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 363 ec->fir_state.coeffs = ec->fir_taps16[0];
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 364 echo_value = fir16(&ec->fir_state, tx);
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 365 ec->clean = rx - echo_value;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 366 ec->lcleanacc += abs(ec->clean) - ec->lclean;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 367 ec->lclean = (ec->lcleanacc + (1 << 4)) >> 5;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 368
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 369 /* Background filter */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 370
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 371 echo_value = fir16(&ec->fir_state_bg, tx);
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 372 clean_bg = rx - echo_value;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 373 ec->lclean_bgacc += abs(clean_bg) - ec->lclean_bg;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 374 ec->lclean_bg = (ec->lclean_bgacc + (1 << 4)) >> 5;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 375
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 376 /* Background Filter adaption */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 377
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 378 /* Almost always adap bg filter, just simple DT and energy
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 379 detection to minimise adaption in cases of strong double talk.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 380 However this is not critical for the dual path algorithm.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 381 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 382 ec->factor = 0;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 383 ec->shift = 0;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 @384 if ((ec->nonupdate_dwell == 0)) {
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 385 int p, logp, shift;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 386
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 387 /* Determine:
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 388
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 389 f = Beta * clean_bg_rx/P ------ (1)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 390
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 391 where P is the total power in the filter states.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 392
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 393 The Boffins have shown that if we obey (1) we converge
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 394 quickly and avoid instability.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 395
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 396 The correct factor f must be in Q30, as this is the fixed
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 397 point format required by the lms_adapt_bg() function,
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 398 therefore the scaled version of (1) is:
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 399
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 400 (2^30) * f = (2^30) * Beta * clean_bg_rx/P
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 401 factor = (2^30) * Beta * clean_bg_rx/P ----- (2)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 402
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 403 We have chosen Beta = 0.25 by experiment, so:
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 404
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 405 factor = (2^30) * (2^-2) * clean_bg_rx/P
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 406
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 407 (30 - 2 - log2(P))
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 408 factor = clean_bg_rx 2 ----- (3)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 409
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 410 To avoid a divide we approximate log2(P) as top_bit(P),
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 411 which returns the position of the highest non-zero bit in
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 412 P. This approximation introduces an error as large as a
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 413 factor of 2, but the algorithm seems to handle it OK.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 414
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 415 Come to think of it a divide may not be a big deal on a
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 416 modern DSP, so its probably worth checking out the cycles
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 417 for a divide versus a top_bit() implementation.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 418 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 419
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 420 p = MIN_TX_POWER_FOR_ADAPTION + ec->pstates;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 421 logp = top_bit(p) + ec->log2taps;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 422 shift = 30 - 2 - logp;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 423 ec->shift = shift;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 424
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 425 lms_adapt_bg(ec, clean_bg, shift);
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 426 }
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 427
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 428 /* very simple DTD to make sure we dont try and adapt with strong
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 429 near end speech */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 430
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 431 ec->adapt = 0;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 432 if ((ec->lrx > MIN_RX_POWER_FOR_ADAPTION) && (ec->lrx > ec->ltx))
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 433 ec->nonupdate_dwell = DTD_HANGOVER;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 434 if (ec->nonupdate_dwell)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 435 ec->nonupdate_dwell--;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 436
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 437 /* Transfer logic */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 438
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 439 /* These conditions are from the dual path paper [1], I messed with
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 440 them a bit to improve performance. */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 441
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 442 if ((ec->adaption_mode & ECHO_CAN_USE_ADAPTION) &&
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 443 (ec->nonupdate_dwell == 0) &&
> > > dc57a3ea80a85a drivers/staging/echo/echo.c Alexander Beregalov 2009-03-12 444 /* (ec->Lclean_bg < 0.875*ec->Lclean) */
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 445 (8 * ec->lclean_bg < 7 * ec->lclean) &&
> > > dc57a3ea80a85a drivers/staging/echo/echo.c Alexander Beregalov 2009-03-12 446 /* (ec->Lclean_bg < 0.125*ec->Ltx) */
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 447 (8 * ec->lclean_bg < ec->ltx)) {
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 448 if (ec->cond_met == 6) {
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 449 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 450 * BG filter has had better results for 6 consecutive
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 451 * samples
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 452 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 453 ec->adapt = 1;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 454 memcpy(ec->fir_taps16[0], ec->fir_taps16[1],
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 455 ec->taps * sizeof(int16_t));
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 456 } else
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 457 ec->cond_met++;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 458 } else
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 459 ec->cond_met = 0;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 460
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 461 /* Non-Linear Processing */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 462
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 463 ec->clean_nlp = ec->clean;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 464 if (ec->adaption_mode & ECHO_CAN_USE_NLP) {
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 465 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 466 * Non-linear processor - a fancy way to say "zap small
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 467 * signals, to avoid residual echo due to (uLaw/ALaw)
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 468 * non-linearity in the channel.".
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 469 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 470
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 471 if ((16 * ec->lclean < ec->ltx)) {
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 472 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 473 * Our e/c has improved echo by at least 24 dB (each
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 474 * factor of 2 is 6dB, so 2*2*2*2=16 is the same as
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 475 * 6+6+6+6=24dB)
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 476 */
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 477 if (ec->adaption_mode & ECHO_CAN_USE_CNG) {
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 478 ec->cng_level = ec->lbgn;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 479
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 480 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 481 * Very elementary comfort noise generation.
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 482 * Just random numbers rolled off very vaguely
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 483 * Hoth-like. DR: This noise doesn't sound
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 484 * quite right to me - I suspect there are some
> > > 83aa3c7bf3f04a drivers/staging/echo/echo.c Jonathan Neuschäfer 2011-03-01 485 * overflow issues in the filtering as it's too
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 486 * "crackly".
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 487 * TODO: debug this, maybe just play noise at
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 488 * high level or look at spectrum.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 489 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 490
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 491 ec->cng_rndnum =
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 492 1664525U * ec->cng_rndnum + 1013904223U;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 493 ec->cng_filter =
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 494 ((ec->cng_rndnum & 0xFFFF) - 32768 +
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 495 5 * ec->cng_filter) >> 3;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 496 ec->clean_nlp =
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 497 (ec->cng_filter * ec->cng_level * 8) >> 14;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 498
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 499 } else if (ec->adaption_mode & ECHO_CAN_USE_CLIP) {
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 500 /* This sounds much better than CNG */
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 501 if (ec->clean_nlp > ec->lbgn)
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 502 ec->clean_nlp = ec->lbgn;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 503 if (ec->clean_nlp < -ec->lbgn)
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 504 ec->clean_nlp = -ec->lbgn;
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 505 } else {
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 506 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 507 * just mute the residual, doesn't sound very
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 508 * good, used mainly in G168 tests
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 509 */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 510 ec->clean_nlp = 0;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 511 }
> > > 4460a860f72898 drivers/staging/echo/echo.c J.R. Mauro 2008-10-20 512 } else {
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 513 /*
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 514 * Background noise estimator. I tried a few
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 515 * algorithms here without much luck. This very simple
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 516 * one seems to work best, we just average the level
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 517 * using a slow (1 sec time const) filter if the
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 518 * current level is less than a (experimentally
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 519 * derived) constant. This means we dont include high
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 520 * level signals like near end speech. When combined
> > > 49bb9e6d753469 drivers/staging/echo/echo.c Greg Kroah-Hartman 2009-08-10 521 * with CNG or especially CLIP seems to work OK.
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 522 */
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 523 if (ec->lclean < 40) {
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 524 ec->lbgn_acc += abs(ec->clean) - ec->lbgn;
> > > 0c474826cfdd24 drivers/staging/echo/echo.c Lisa Nguyen 2013-05-05 525 ec->lbgn = (ec->lbgn_acc + (1 << 11)) >> 12;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 526 }
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 527 }
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 528 }
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 529
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 530 /* Roll around the taps buffer */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 531 if (ec->curr_pos <= 0)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 532 ec->curr_pos = ec->taps;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 533 ec->curr_pos--;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 534
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 535 if (ec->adaption_mode & ECHO_CAN_DISABLE)
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 536 ec->clean_nlp = rx;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 537
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 538 /* Output scaled back up again to match input scaling */
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 539
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 540 return (int16_t) ec->clean_nlp << 1;
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 541 }
> > > 9d8f2d5dfbcc9d drivers/staging/echo/echo.c Tzafrir Cohen 2008-10-12 542 EXPORT_SYMBOL_GPL(oslec_update);
> > > 10602db812fa27 drivers/staging/echo/echo.c David Rowe 2008-10-06 543
> > >
> > > :::::: The code at line 384 was first introduced by commit
> > > :::::: 10602db812fa270fc923f5e48fb47202288828f3 Staging: add echo cancelation module
> > >
> > > :::::: TO: David Rowe <david@xxxxxxxxxxx>
> > > :::::: CC: Greg Kroah-Hartman <gregkh@xxxxxxx>
> > >
> > > ---
> > > 0-DAY CI Kernel Test Service, Intel Corporation
> > > https://lists.01.org/hyperkitty/list/kbuild-all@xxxxxxxxxxxx
> > >
--
Thanks,
~Nick Desaulniers
