From randress@west.raytheon.com  Fri Feb  5 11:26:57 1999
From: randress@west.raytheon.com
Subject: ACT #273, Clobberd by longjmp or vfork

I am unable to find documentation on these warnings except page 23 of Using and
Porting GNU CC? Is there more? What do they mean when compiling GNAT Ada95?

some_package.adb: In function `some_function':
some_package.adb:763: warning: variable `var1' might be clobbered by `longjmp'
or `vfork'
some_package.adb:784: warning: variable `var2' might be clobbered by `longjmp'
or `vfork'
some_package.adb:787: warning: variable `var2' might be clobbered by `longjmp'
or `vfork'


Randal Andress


From report@gnat.com
 Re: [6205-002] - ACT #273, Clobberd by longjmp or vfork

 
 [Automatic message from GNAT report tracking mechanism]
 
 Dear Customer,
 
 Your report has been automatically assigned tracking number [6205-002].
 Please be sure to put this number in the subject line of any
 subsequent followup message. Messages from us on this problem will
 quote this tracking number.
 
 Thank you for using GNAT.
 
 -- The GNAT support team

From dewar@gnat.com  Fri Feb  5 11:32:32 1999
  Re: [6205-002] - ACT #273, Clobberd by longjmp or vfork

 these messages can be ignored, they have no effect on the generated code.
 They will disappear eventually when we complete the implementation of
 fast exception support on all targets.

From randress@west.raytheon.com  Fri Feb  5 12:07:14 1999
  Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfork

 
 Thank-you. We will ignore them, as I do not see any way to turn them off without
 loosing other more useful warnings.
 
 I was wondering about 'fast exception support' implementation. Is this the
 'table' vs 'stack' based scheme?  I infer from your message that the MIPS/IRIX
 implementation (3.11b, 3.12w) still uses the stack based approach - fast in
 execution of exceptions taken/raised but slower to execute than the table driven
 method if exceptions do not occur. Would you please confirm or correct these
 perceptions?
 
 Thanks again for your prompt attention to this report.
 
 Randal Andress
 
 ____________________Reply Separator____________________
 Subject:    Re: [6205-002] - ACT #273, Clobberd by longjmp or vfork
 Author: dewar@gnat.com
 Date:       2/5/99 8:13 AM
 
 these messages can be ignored, they have no effect on the generated code.
 They will disappear eventually when we complete the implementation of
 fast exception support on all targets.
 RFC-822-headers:
 Received: from CONVERSION-DAEMON by mail.hac.com (PMDF V5.1-12 #26580)
  id <0F6O00301VQAAM@mail.hac.com>; Fri,  5 Feb 1999 08:13:26 -0800 (PST)
 Received: from PROCESS-DAEMON by mail.hac.com (PMDF V5.1-12 #26580)
  id <0F6O00301VQ99N@mail.hac.com>; Fri, 05 Feb 1999 08:13:22 -0800 (PST)
 Received: from fw-tu05.hac.com by mail.hac.com (PMDF V5.1-12 #26580)
  with ESMTP id <0F6O0024UVQ7NO@mail.hac.com>; Fri,
  05 Feb 1999 08:13:20 -0800 (PST)
 Received: from nile.gnat.com ([205.232.38.5]) by fw-tu05.hac.com (8.9.0/8.9.0)
  with SMTP id JAA09246; Fri, 05 Feb 1999 09:15:59 -0700 (MST)
 Received: by nile.gnat.com (5.0/1.20) id AA14682; Fri,
  05 Feb 1999 11:13:54 -0500 (EST)
 Date: Fri, 05 Feb 1999 11:13:54 -0500 (EST)
 From: dewar@gnat.com (Robert Dewar)
 Subject: Re: [6205-002] - ACT #273, Clobberd by longjmp or vfork
 Message-id: <9902051613.AA14682@nile.gnat.com>
 MIME-version: 1.0
 

From charlet@ACT-Europe.FR  Fri Feb  5 12:14:21 1999
  Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfork

 > Thank-you. We will ignore them, as I do not see any way to turn them off without
 > loosing other more useful warnings.
 
 That's the best thing to do indeed.
 You night want to try to put pragma Warnings (On/Off) at very localized places,
 but it might be easy.
 
 > I was wondering about 'fast exception support' implementation. Is this the
 > 'table' vs 'stack' based scheme?  I infer from your message that the MIPS/IRIX
 > implementation (3.11b, 3.12w) still uses the stack based approach - fast in
 > execution of exceptions taken/raised but slower to execute than the table driven
 > method if exceptions do not occur. Would you please confirm or correct these
 > perceptions?
 
 Your perceptions are correct. Note that we intend to make the table based
 scheme the default for the next version of GNAT (3.12a) on targets where it
 is supported.
 
 Regards,
 
 Arno

From charlet@ACT-Europe.FR  Fri Feb  5 13:38:03 1999
  Re: Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or

 > One more question before we close this one...
 > When do you anticipate it will be supported for MIPS/IRIX?
 
 It is already supported for MIPS/IRIX, it is just not the default.
 If you want to enable this capability you need to recompile everything 
 (even the GNAT run time) with the -gnatdX flag, e.g using
 
 gnatmake -f -a -gnatdX
 
 Regards,
 
 Arno

From randress@west.raytheon.com  Fri Feb  5 13:38:36 1999
  Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or

 
 One more question before we close this one...
 When do you anticipate it will be supported for MIPS/IRIX?
 
 Thanks,
 -Randal
 
 ____________________Reply Separator____________________
 Subject:    Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfo
 Author: charlet@ACT-Europe.FR
 Date:       2/5/99 8:57 AM
 
 > Thank-you. We will ignore them, as I do not see any way to turn them off
    without
 > loosing other more useful warnings.
 
 That's the best thing to do indeed.
 You night want to try to put pragma Warnings (On/Off) at very localized places,
 but it might be easy.
 
 > I was wondering about 'fast exception support' implementation. Is this the
 > 'table' vs 'stack' based scheme?  I infer from your message that the MIPS/IRIX
 > implementation (3.11b, 3.12w) still uses the stack based approach - fast in
 > execution of exceptions taken/raised but slower to execute than the table
    driven
 > method if exceptions do not occur. Would you please confirm or correct these
 > perceptions?
 
 Your perceptions are correct. Note that we intend to make the table based
 scheme the default for the next version of GNAT (3.12a) on targets where it
 is supported.
 
 Regards,
 
 Arno
 RFC-822-headers:
 Received: from CONVERSION-DAEMON by mail.hac.com (PMDF V5.1-12 #26580)
  id <0F6O00H01XNVOW@mail.hac.com>; Fri,  5 Feb 1999 08:55:10 -0800 (PST)
 Received: from PROCESS-DAEMON by mail.hac.com (PMDF V5.1-12 #26580)
  id <0F6O00H01XNTOM@mail.hac.com>; Fri, 05 Feb 1999 08:55:06 -0800 (PST)
 Received: from fw-es05.hac.com by mail.hac.com (PMDF V5.1-12 #26580)
  with ESMTP id <0F6O00DJCXNTUE@mail.hac.com>; Fri,
  05 Feb 1999 08:55:05 -0800 (PST)
 Received: from paris.act-europe.fr ([195.25.32.140])
  by fw-es05.hac.com (8.9.0/8.9.0) with ESMTP id IAA01840; Fri,
  05 Feb 1999 08:57:42 -0800 (PST)
 Received: from dublin.act-europe.fr (dublin.act-europe.fr [195.25.32.154])
  by paris.act-europe.fr (8.9.1a/8.9.1) with ESMTP id RAA06330; Fri,
  05 Feb 1999 17:57:10 +0100
 Received: (from charlet@localhost) by dublin.act-europe.fr (8.9.1a/8.9.1)
  id RAA25529; Fri, 05 Feb 1999 17:57:09 +0100 (MET)
 Date: Fri, 05 Feb 1999 17:57:09 +0100 (MET)
 From: Arnaud Charlet <charlet@ACT-Europe.FR>
 Subject: Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfork
 In-reply-to: <0F6O00I1SXK6F7@mail.hac.com> from
  <"randress@west.raytheon.com"@Feb>
 Message-id: <199902051657.RAA25529@dublin.act-europe.fr>
 MIME-version: 1.0
 X-Mailer: ELM [version 2.4 PL20]
 Content-transfer-encoding: 7BIT
 

From randress@west.raytheon.com  Fri Feb  5 14:39:32 1999
  Re[2]: Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by long

 
 Thank you,
 
 We will probably wait for you to do this in a delivered/default version...but
 thanks for the clear, prompt reply.
 
 -Randal
 
 ____________________Reply Separator____________________
 Subject:    Re: Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp
 Author: charlet@ACT-Europe.FR
 Date:       2/5/99 10:20 AM
 
 > One more question before we close this one...
 > When do you anticipate it will be supported for MIPS/IRIX?
 
 It is already supported for MIPS/IRIX, it is just not the default.
 If you want to enable this capability you need to recompile everything
 (even the GNAT run time) with the -gnatdX flag, e.g using
 
 gnatmake -f -a -gnatdX
 
 Regards,
 
 Arno
 RFC-822-headers:
 Received: from CONVERSION-DAEMON by mail.hac.com (PMDF V5.1-12 #D3246)
  id <0F6P00L011RC15@mail.hac.com>; Fri,  5 Feb 1999 10:23:38 -0800 (PST)
 Received: from PROCESS-DAEMON by mail.hac.com (PMDF V5.1-12 #D3246)
  id <0F6P00L011RB0W@mail.hac.com>; Fri, 05 Feb 1999 10:23:36 -0800 (PST)
 Received: from fw-es05.hac.com by mail.hac.com (PMDF V5.1-12 #D3246)
  with ESMTP id <0F6P00K3L1RA0V@mail.hac.com>; Fri,
  05 Feb 1999 10:23:34 -0800 (PST)
 Received: from paris.act-europe.fr ([195.25.32.140])
  by fw-es05.hac.com (8.9.0/8.9.0) with ESMTP id KAA16081; Fri,
  05 Feb 1999 10:21:06 -0800 (PST)
 Received: from dublin.act-europe.fr (dublin.act-europe.fr [195.25.32.154])
  by paris.act-europe.fr (8.9.1a/8.9.1) with ESMTP id TAA08198; Fri,
  05 Feb 1999 19:20:34 +0100
 Received: (from charlet@localhost) by dublin.act-europe.fr (8.9.1a/8.9.1)
  id TAA25034; Fri, 05 Feb 1999 19:20:34 +0100 (MET)
 Date: Fri, 05 Feb 1999 19:20:34 +0100 (MET)
 From: Arnaud Charlet <charlet@ACT-Europe.FR>
 Subject: Re: Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or
 In-reply-to: <0F6P00K1219UHU@mail.hac.com> from
  <"randress@west.raytheon.com"@Feb>
 Message-id: <199902051820.TAA25034@dublin.act-europe.fr>
 MIME-version: 1.0
 X-Mailer: ELM [version 2.4 PL20]
 Content-transfer-encoding: 7BIT
 

From dewar@gnat.com  Fri Feb  5 15:48:47 1999
  Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfork

 <<That's the best thing to do indeed.
 You night want to try to put pragma Warnings (On/Off) at very localized places,
 but it might be easy.
 >>
 
 That will not work, these are gcc warnings, not gnat front end warnings, the
 pragma affects only the latter.

From randress@west.raytheon.com  Fri Feb  5 16:42:31 1999
  Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or

 
 Thank-you for the clarification....I haven't absorbed all the GCC warning
 options, but I assume that if there were a way to turn just these warnings off,
 you would be quite familiar with it...Not so much that this is a problem, we
 just wanted to understand what the implications were - finding none harmful, we
 can ignore them for the moment.
 
 Thanks again
 Randal
 
 ____________________Reply Separator____________________
 Subject:    Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfo
 Author: dewar@gnat.com
 Date:       2/5/99 12:30 PM
 
 <<That's the best thing to do indeed.
 You night want to try to put pragma Warnings (On/Off) at very localized places,
 but it might be easy.
 >>
 
 That will not work, these are gcc warnings, not gnat front end warnings, the
 pragma affects only the latter.

From randress@west.raytheon.com Fri Feb  5 11:46 EST 1999
Subject: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfork
To: dewar@gnat.com, report@gnat.com
Cc: djbergquist@west.raytheon.com, ebruce@west.raytheon.com,
        jfbratton@west.raytheon.com, jhamilton@west.raytheon.com

 Thank-you. We will ignore them, as I do not see any way to turn them off without
 loosing other more useful warnings.
 
 I was wondering about 'fast exception support' implementation. Is this the
 'table' vs 'stack' based scheme?  I infer from your message that the MIPS/IRIX
 implementation (3.11b, 3.12w) still uses the stack based approach - fast in
 execution of exceptions taken/raised but slower to execute than the table driven
 method if exceptions do not occur. Would you please confirm or correct these
 perceptions?

From dewar@gnat.com Sun Feb  7 07:38 EST 1999
From: dewar@gnat.com (Robert Dewar)
To: dewar@gnat.com, randress@west.raytheon.com, report@gnat.com
Cc: djbergquist@west.raytheon.com, ebruce@west.raytheon.com,
        jfbratton@west.raytheon.com, jhamilton@west.raytheon.com

 You asked

 <<I was wondering about 'fast exception support' implementation. Is this the
 'table' vs 'stack' based scheme?  I infer from your message that the MIPS/IRIX
 implementation (3.11b, 3.12w) still uses the stack based approach - fast in
 execution of exceptions taken/raised but slower to execute than the table driven
 method if exceptions do not occur. Would you please confirm or correct these
 perceptions?>>

 The MIPS IRIX implementation of GNAT from SGI uses only fast exceptions.
 The version from us allows either method, with slow exceptions being
 the default. The fast exceptions can be enabled by
 
 gnatmake -f -a -gnatdX program

From randress@west.raytheon.com  Mon Feb  8 15:04:59 1999
  Re[2]: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or

 
 Thanks for the clarification..
 Randal Andress
 
 ____________________Reply Separator____________________
 Subject:    Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfo
 Author: dewar@gnat.com
 Date:       2/7/99 4:38 AM
 
 You asked
 
 <<I was wondering about 'fast exception support' implementation. Is this the
 'table' vs 'stack' based scheme?  I infer from your message that the MIPS/IRIX
 implementation (3.11b, 3.12w) still uses the stack based approach - fast in
 execution of exceptions taken/raised but slower to execute than the table driven
 method if exceptions do not occur. Would you please confirm or correct these
 perceptions?
 >>
 
 The MIPS IRIX implementation of GNAT from SGI uses only fast exceptions.
 The version from us allows either method, with slow exceptions being
 the default. The fast exceptions can be enabled by
 
 gnatmake -f -a -gnatdX program
 RFC-822-headers:
 Received: from CONVERSION-DAEMON by mail.hac.com (PMDF V5.1-12 #26580)
  id <0F6S00M01B2NWQ@mail.hac.com>; Sun,  7 Feb 1999 04:37:36 -0800 (PST)
 Received: from PROCESS-DAEMON by mail.hac.com (PMDF V5.1-12 #26580)
  id <0F6S00M01B2MWP@mail.hac.com>; Sun, 07 Feb 1999 04:37:35 -0800 (PST)
 Received: from fw-es05.hac.com by mail.hac.com (PMDF V5.1-12 #26580)
  with ESMTP id <0F6S00M0FB2LMG@mail.hac.com>; Sun,
  07 Feb 1999 04:37:34 -0800 (PST)
 Received: from nile.gnat.com ([205.232.38.5]) by fw-es05.hac.com (8.9.0/8.9.0)
  with SMTP id EAA25700; Sun, 07 Feb 1999 04:40:15 -0800 (PST)
 Received: by nile.gnat.com (5.0/1.20) id AA04351; Sun,
  07 Feb 1999 07:38:05 -0500 (EST)
 Date: Sun, 07 Feb 1999 07:38:05 -0500 (EST)
 From: dewar@gnat.com (Robert Dewar)
 Subject: Re: Re[2]: [6205-002] - ACT #273, Clobberd by longjmp or vfork
 Message-id: <9902071238.AA04351@nile.gnat.com>
 MIME-version: 1.0
 

