public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Szabolcs Nagy <szabolcs.nagy@arm.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>,
	libc-alpha@sourceware.org, Bruno Haible <bruno@clisp.org>
Subject: Re: [PATCH 5/7] riscv: Fix feenvupdate with FE_DFL_ENV (BZ 31022)
Date: Tue, 9 Jan 2024 13:44:06 +0000	[thread overview]
Message-ID: <ZZ1NpmWCqQ+HZSoV@arm.com> (raw)
In-Reply-To: <20231103122416.2724355-6-adhemerval.zanella@linaro.org>

The 11/03/2023 09:24, Adhemerval Zanella wrote:
> libc_feupdateenv_riscv should check for FE_DFL_ENV, similar to
> libc_fesetenv_riscv.
> 
> Also extend the test-fenv.c to test fenvupdate.
> 
> Checked on riscv under qemu-system.

the test fails on aarch64 when trapping is supported.

> +++ b/math/test-fenv.c
> @@ -196,6 +196,30 @@ set_single_exc (const char *test_name, int fe_exc, fexcept_t exception)
>    feclearexcept (exception);
>    test_exceptions (str, ALL_EXC ^ fe_exc, 0);
>  }
> +
> +static void
> +update_single_exc (const char *test_name, const fenv_t *envp, int fe_exc,
> +		   int fe_exc_clear, fexcept_t exception)
> +{
> +  char str[200];
> +  /* The standard allows the inexact exception to be set together with the
> +     underflow and overflow exceptions.  So ignore the inexact flag if the
> +     others are raised.  */
> +  int ignore_inexact = (fe_exc & (UNDERFLOW_EXC | OVERFLOW_EXC)) != 0;
> +
> +  strcpy (str, test_name);
> +  strcat (str, ": set flag, with rest not set");
> +  feclearexcept (FE_ALL_EXCEPT);
> +  feraiseexcept (exception);

this relies on disabled trapping.

> +  feupdateenv (envp);
> +  test_exceptions (str, fe_exc, ignore_inexact);
> +
> +  strcpy (str, test_name);
> +  strcat (str, ": clear flag, rest also unset");
> +  feclearexcept (exception);
> +  feupdateenv (envp);
> +  test_exceptions (str, fe_exc_clear, ignore_inexact);
> +}
...
>  /* Test that program aborts with no masked interrupts */
>  static void
> -feenv_nomask_test (const char *flag_name, int fe_exc)
> +feenv_nomask_test (const char *flag_name, int fe_exc, int (*func)(const fenv_t *))
>  {
>  # if defined FE_NOMASK_ENV
>    int status;
>    pid_t pid;
>  
>    if (!EXCEPTION_ENABLE_SUPPORTED (FE_ALL_EXCEPT)
> -      && fesetenv (FE_NOMASK_ENV) != 0)
> +      && func (FE_NOMASK_ENV) != 0)

trapping is enabled here on targets where trapping is optional.

> +static void
> +feupdate_single_test (const char *flag_name, int fe_exc)
> +{
> +  feenv_nomask_test (flag_name, fe_exc, feupdateenv);
> +  feenv_mask_test (flag_name, fe_exc, feupdateenv);
> +}

trapping is still on after this function.

> +static void
> +feupdateenv_single_test (const char *test_name, int fe_exc,
> +			 fexcept_t exception)
> +{
> +  char str[100];
> +  fenv_t env;
> +  int res;
> +
> +  snprintf (str, sizeof str, "feupdateenv %s and FL_DFL_ENV", test_name);
> +  update_single_exc (str, FE_DFL_ENV, fe_exc, NO_EXC, exception);

this relies on disabled trapping.

> +static void
> +feupdateenv_tests (void)
> +{
...
> +#ifdef FE_OVERFLOW
> +  feupdate_single_test ("FE_OVERFLOW", FE_OVERFLOW);
> +#endif
> +
> +#ifdef FE_DIVBYZERO
> +  feupdateenv_single_test ("DIVBYZERO", DIVBYZERO_EXC, FE_DIVBYZERO);
> +#endif

this test fails because previous enables trapping mode.

fwiw a feupdateenv (FE_DFL_ENV); in between the tests
makes math/test-fenv pass.

  parent reply	other threads:[~2024-01-09 13:44 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-11-03 12:24 [PATCH 0/7] Multiple floating-point environment fixes Adhemerval Zanella
2023-11-03 12:24 ` [PATCH 1/7] powerpc: Do not raise exception traps for fesetexcept/fesetexceptflag (BZ 30988) Adhemerval Zanella
2023-11-03 12:24 ` [PATCH 2/7] i686: Do not raise exception traps on fesetexcept (BZ 30989) Adhemerval Zanella
2023-11-06  0:15   ` Bruno Haible
2023-11-06 11:24     ` Adhemerval Zanella Netto
2023-11-03 12:24 ` [PATCH 3/7] x86: Do not raises floating-point exception traps on fesetexceptflag (BZ 30990) Adhemerval Zanella
2023-11-06  0:17   ` Bruno Haible
2023-11-06 11:24     ` Adhemerval Zanella Netto
2023-11-03 12:24 ` [PATCH 4/7] manual: Clarify undefined behavior of feenableexcept (BZ 31019) Adhemerval Zanella
2023-11-03 16:20   ` Joseph Myers
2023-11-06  0:17     ` Bruno Haible
2023-11-06 11:25       ` Adhemerval Zanella Netto
2023-11-03 12:24 ` [PATCH 5/7] riscv: Fix feenvupdate with FE_DFL_ENV (BZ 31022) Adhemerval Zanella
2023-11-06  0:21   ` Bruno Haible
2023-11-06 11:28     ` Adhemerval Zanella Netto
2023-11-06  0:24   ` Bruno Haible
2023-11-06 11:32     ` Adhemerval Zanella Netto
2024-01-09 13:44   ` Szabolcs Nagy [this message]
2024-01-09 18:16     ` Adhemerval Zanella Netto
2023-11-03 12:24 ` [PATCH 6/7] alpha: Fix fesetexceptflag (BZ 30998) Adhemerval Zanella
2023-11-03 12:24 ` [PATCH 7/7] hppa: Fix undefined behaviour in feclearexcept (BZ 30983) Adhemerval Zanella
2023-11-06  0:25 ` [PATCH 0/7] Multiple floating-point environment fixes Bruno Haible

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=ZZ1NpmWCqQ+HZSoV@arm.com \
    --to=szabolcs.nagy@arm.com \
    --cc=adhemerval.zanella@linaro.org \
    --cc=bruno@clisp.org \
    --cc=libc-alpha@sourceware.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).