RE: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

Hi, > If you have access to x86_64-linux, you can easily test it yourself with: > make -C gcc check-gcc RUNTESTFLAGS='--target_board=unix\{-m64,-m32/- > march=i386\} dg.exp=pr61441.c' > > Your patch does help. Thanks, I tested the changes and it worked fine. > issignalling is a GNU extension

Re: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

On Thu, Jan 07, 2016 at 06:48:25AM +, Saraswati, Sujoy (OSTL) wrote: > Hi, > > > On Thu, Nov 26, 2015 at 08:38:55AM +, Saraswati, Sujoy (OSTL) wrote: > > > PR tree-optimization/61441 > > > * gcc.dg/pr61441.c: New testcase. > > > > Note the testcase fails on

RE: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

Hi, > On Thu, Nov 26, 2015 at 08:38:55AM +, Saraswati, Sujoy (OSTL) wrote: > > PR tree-optimization/61441 > > * gcc.dg/pr61441.c: New testcase. > > Note the testcase fails on i686-linux, and even -fexcess-precision=standard > doesn't help (-ffloat-store works, but

Re: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

On Thu, Nov 26, 2015 at 08:38:55AM +, Saraswati, Sujoy (OSTL) wrote: > PR tree-optimization/61441 > * gcc.dg/pr61441.c: New testcase. Note the testcase fails on i686-linux, and even -fexcess-precision=standard doesn't help (-ffloat-store works, but that is a big hammer

RE: Fix 61441 [ 1/5] Add REAL_VALUE_ISSIGNALING_NAN

Hi, > If you haven't set up write-access to the repository, please go ahead and get > that process started: > > https://www.gnu.org/software/gcc/svnwrite.html > > You can list me as your sponsor on the form. > > Once your account is set up, you can commit patches which have been > approved. >

Re: Fix 61441 [ 1/5] Add REAL_VALUE_ISSIGNALING_NAN

On 11/26/2015 01:27 AM, Saraswati, Sujoy (OSTL) wrote: Hi, This series of patches fixes PR61441. The fix is broken into 5 patches. The first one adds REAL_VALUE_ISSIGNALING_NAN. 2015-11-26 Sujoy Saraswati PR tree-optimization/61441 * real.c

Re: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

On 11/26/2015 01:38 AM, Saraswati, Sujoy (OSTL) wrote: Hi, This patch avoids various transformations with signaling NaN operands when flag_signaling_nans is on, to avoid folding which would lose exceptions. A test case for this change is also added as part of this patch. Regards, Sujoy

Re: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

On Wed, 2 Dec 2015, Jeff Law wrote: > Why let NEGATE_EXPR and ABS_EXPR pass through here? I realize that these can > often be implemented with bit-twiddling, so they're usually considered > special. BUt in this case aren't we just dealing with constants and wouldn't > we want to still express

Re: Fix 61441 [5/5] Disable various transformations for signaling NaN operands

On 12/02/2015 12:08 PM, Joseph Myers wrote: On Wed, 2 Dec 2015, Jeff Law wrote: Why let NEGATE_EXPR and ABS_EXPR pass through here? I realize that these can often be implemented with bit-twiddling, so they're usually considered special. BUt in this case aren't we just dealing with constants

RE: Fix 61441

Hi, > I think the general principle is: > > * The caller decides whether folding is desirable (whether it would lose > exceptions, for example). > > * The real.c code is called only when the caller has decided that folding is > desirable, and should always produce the correct output (which for

Fix 61441 [ 1/5] Add REAL_VALUE_ISSIGNALING_NAN

Hi, This series of patches fixes PR61441. The fix is broken into 5 patches. The first one adds REAL_VALUE_ISSIGNALING_NAN. 2015-11-26 Sujoy Saraswati PR tree-optimization/61441 * real.c (real_issignaling_nan): New. * real.h

Fix 61441 [2/5] Use REAL_VALUE_ISSIGNALING_NAN instead of REAL_VALUE_ISNAN where appropriate

This patch uses REAL_VALUE_ISSIGNALING_NAN instead of REAL_VALUE_ISNAN to avoid the operation for sNaN. Regards, Sujoy 2015-11-26 Sujoy Saraswati PR tree-optimization/61441 * fold-const.c (const_binop): Use REAL_VALUE_ISSIGNALING_NAN

Fix 61441 [3/5] Remove flag_errno_math check for RINT

Hi, This patch removes flag_errno_math check for RINT, treating it similar to nearbyint. Regards, Sujoy 2015-11-26 Sujoy Saraswati PR tree-optimization/61441 * match.pd (f(x) -> x): Removed flag_errno_math check for RINT. Index:

Fix 61441 [4/5] Produce quiet NaN for real value operations

Hi, This patch makes resulting NaN values to be quiet NaN for real value operations, irrespective of the flag_signaling_nans flag. The caller has the responsibility to avoid the operation if flag_signaling_nans is on. Regards, Sujoy 2015-11-26 Sujoy Saraswati

Fix 61441 [5/5] Disable various transformations for signaling NaN operands

Hi, This patch avoids various transformations with signaling NaN operands when flag_signaling_nans is on, to avoid folding which would lose exceptions. A test case for this change is also added as part of this patch. Regards, Sujoy 2015-11-26 Sujoy Saraswati

Re: Fix 61441 [3/5] Remove flag_errno_math check for RINT

On Thu, Nov 26, 2015 at 9:31 AM, Saraswati, Sujoy (OSTL) wrote: > Hi, > This patch removes flag_errno_math check for RINT, treating it similar to > nearbyint. > Regards, > Sujoy Ok. Richard. > 2015-11-26 Sujoy Saraswati > > PR

Re: Fix 61441

On Fri, 6 Nov 2015, Sujoy Saraswati wrote: > > Shouldn't real_convert do this rather than the caller needing to do it? > > Yes, it should be. I had started by doing this within real_convert but > then saw that there are quite a few callers where I should add the > check for flag_signaling_nans.

Re: Fix 61441

Hi, > It seems this code has moved to fold-const.c, so the patch will need > updating. I have updated the patch and sending again. > Now, is it actually correct for integer_valued_real_single_p (now) to > treat sNaN as an integer? Looking at existing uses, they all appear to be > in match.pd:

Re: Fix 61441

On Thu, 5 Nov 2015, Sujoy Saraswati wrote: > > Some other places in this patch have similar issues with checking > > HONOR_SNANS && REAL_VALUE_ISNAN when they should check if the particular > > value is sNaN. > > I modified the code to check for sNaN specifically. The modified patch > is below.

Re: Fix 61441

Hi, > This is looking pretty close to ready to go in, but the patch also seems > to be doing several things at once and it might be easier to review if > split into separate logical pieces. For example (this is an illustration, > not necessarily saying it should be exactly this way): (1) Add >

Re: Fix 61441

On Tue, 13 Oct 2015, Sujoy Saraswati wrote: > Index: gcc/builtins.c > === > --- gcc/builtins.c (revision 228700) > +++ gcc/builtins.c (working copy) > @@ -7357,7 +7357,11 @@ integer_valued_real_p (tree t) > &&

Re: [Ping] Fix 61441

This is a ping for the patch to fix 61441. Regards, Sujoy On Tue, Oct 13, 2015 at 4:16 PM, Sujoy Saraswati <ssarasw...@gmail.com> wrote: > Hi, > This is another modified version of the patch, incorporating the > previous comments. > > Bootstrap and regression test

Re: Fix 61441

Hi, This is another modified version of the patch, incorporating the previous comments. Bootstrap and regression tests on x86_64-linux-gnu and aarch64-unknown-linux-gnu passed with changes done on trunk. Is this fine ? Regards, Sujoy 2015-10-13 Sujoy Saraswati

Re: Fix 61441

Hi, >> I'll leave the correctness part of the patch to Joseph who knows FP >> arithmetic better than me, >> implementation-wise this is ok if you fix the REAL_CST sharing issue. Ok, will change this. > Changing fold_abs_const is wrong - abs of sNaN is sNaN, no exceptions > raised. Changing

Re: Fix 61441

On Wed, 16 Sep 2015, Sujoy Saraswati wrote: > > If -fsignaling-nans, then folding of expressions involving sNaNs should be > > disabled, outside of static initializers - such expressions should not get > > folded to return an sNaN (it's incorrect to fold sNaN + 1 to sNaN, for > > example). I

Re: Fix 61441

On Thu, Sep 10, 2015 at 9:30 AM, Sujoy Saraswati wrote: > Hi, > Here is a modified patch for this. The change this time is in > fold-const.c and real.c. > > Bootstrap and regression tests on x86_64-linux-gnu and > aarch64-unknown-linux-gnu passed with changes done on trunk.

Re: Fix 61441

On Mon, 14 Sep 2015, Richard Biener wrote: > I'll leave the correctness part of the patch to Joseph who knows FP > arithmetic better than me, > implementation-wise this is ok if you fix the REAL_CST sharing issue. Changing fold_abs_const is wrong - abs of sNaN is sNaN, no exceptions raised.

Re: Fix 61441

Hi, Here is a modified patch for this. The change this time is in fold-const.c and real.c. Bootstrap and regression tests on x86_64-linux-gnu and aarch64-unknown-linux-gnu passed with changes done on trunk. Is this fine ? Regards, Sujoy 2015-09-10 Sujoy Saraswati

Re: Fix 61441

Hi Richard, > Note that I'm curious what > the actual bug is - is it that (double) sNaN creates a sNaN? Then the fix > should be elsewhere, in constant folding itself > (fold_convert_const_real_from_real > or real_convert). > > If that isn't the bug you have very many other passes to fix for the

Re: Fix 61441

On Wed, Sep 2, 2015 at 1:36 PM, Sujoy Saraswati wrote: > Hi Richard, > >> Note that I'm curious what >> the actual bug is - is it that (double) sNaN creates a sNaN? Then the fix >> should be elsewhere, in constant folding itself >> (fold_convert_const_real_from_real >> or

Re: Fix 61441

> CCP and other passes ultimatively end up using fold-const.c:const_{unop,binop} > for constant folding so that is where the fix should go to (or to real.c). > That > will automatically handle other passes doing similar transforms. Thanks for the tip. I will modify my fix and post it. Regards,

Fix 61441

The following patch fixes 61441. It converts sNaN to qNaN on folding when -fno-signaling-nans is used. Bootstrap and regression tests on x86_64-linux-gnu and aarch64-unknown-linux-gnu passed with changes done on trunk. Is this fix fine ? Regards, Sujoy 2015-09-01 Sujoy Saraswati

Re: Fix 61441

On Tue, Sep 1, 2015 at 12:23 PM, Sujoy Saraswati wrote: > The following patch fixes 61441. It converts sNaN to qNaN on folding > when -fno-signaling-nans is used. > > Bootstrap and regression tests on x86_64-linux-gnu and > aarch64-unknown-linux-gnu passed with changes done