From: Jonathan Wakely <jwakely.gcc@gmail.com>
To: Jonny Grant <jg@jguk.org>
Cc: Xi Ruoyao <xry111@xry111.site>, gcc-help <gcc-help@gcc.gnu.org>
Subject: Re: std::string add nullptr attribute
Date: Mon, 13 Mar 2023 10:10:58 +0000 [thread overview]
Message-ID: <CAH6eHdRkKWPCToJ+HoeiDJg2N2hCqz45c4Z1LUwG3ipaxNLBWw@mail.gmail.com> (raw)
In-Reply-To: <d7536cbb-ec12-f48a-0061-43aaa9462466@jguk.org>
On Sun, 12 Mar 2023 at 22:10, Jonny Grant wrote:
>
>
>
> On 09/02/2023 17:52, Jonathan Wakely wrote:
> > On Thu, 9 Feb 2023 at 16:30, Xi Ruoyao wrote:
> >>
> >> On Thu, 2023-02-09 at 14:56 +0000, Jonathan Wakely via Gcc-help wrote:
> >>>> Note, my code isn't like this, it is just an example to suggest
> >>>> adding the nullptr attribute, as its clearly already rejected at
> >>>> runtime.
> >>>
> >>> I assume you mean the nonnull attribute. That was added in 2020 and
> >>> then reverted because it broke some things:
> >>
> >> I remember I'd once made the same mistake when I suggested to add
> >> nonnull for ostream::operator<<(const string &) and I was lectured:
> >> nonnull is not only a diagnostic attribute, it also allows the compiler
> >> to assume the parameter is never null and rendering std::string(nullptr)
> >> an undefined behavior.
> >
> > Yes, I think that's what might have happened with the std::string change.
>
> How about adding a method that is called by these operators that has the __attribute__ ((nonnull)) ?
>
> example:
> https://godbolt.org/z/bqW86PP34
>
> >> Then the example may just silently continue to run, instead of throwing
> >> an exception. It would be an ironic example: an attempt to improve
> >> diagnostic finally made diagnostic more difficult.
> >
> > Indeed.
> >
> > Maybe we can add __attribute__((access(read, 1))) instead, which says
> > that we will read from the pointer, which also implies it must be
> > non-null.
> >
> > N.B. in C++23 string(nullptr) produces an error, although
> > string((const char*)nullptr) doesn't, so in practice it only prevents
> > the dumbest calls with a literal 'nullptr' token, and not the more
> > realistic problems where you have a pointer that happens to be null.
>
> There is a way to generate a build error for even string((const char*)nullptr)
>
> I made another example that detects nullptr being passed around (should such stupid code occur) at build time providing optimizer is on. With -O0 it just gives the error always; so I put in an __OPTIMIZE__ check. This example doesn't need -fanalyzer.
>
> https://godbolt.org/z/TdGnno4K5
>
> #if __OPTIMIZE__
> void nullptr_compile_abort() __attribute__((error("nullptr compile error")));
> #endif
>
> static void f2(const char * str)
> {
> #if __OPTIMIZE__
> if (str == nullptr) nullptr_compile_abort();
> #endif
> }
>
> int main()
> {
> f2((const char *)nullptr);
> }
This causes compilation to fail for code which is never executed at
run-time, which is not permitted by the standard.
You can use __attribute__((warning(""))) instead, but that is broken^W
inconvenient for inline functions. You need a non-inline definition of
the function, which means exporting a new function from the shared
library just for this diagnostic.
All these techniques you're rediscovering have been tried before :-)
next prev parent reply other threads:[~2023-03-13 10:11 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-09 13:26 Jonny Grant
2023-02-09 14:56 ` Jonathan Wakely
2023-02-09 16:30 ` Xi Ruoyao
2023-02-09 17:52 ` Jonathan Wakely
2023-02-10 21:30 ` Jonny Grant
2023-02-10 22:03 ` Jonathan Wakely
2023-02-10 22:38 ` Jonny Grant
2023-02-11 0:32 ` Jonathan Wakely
2023-02-13 22:02 ` Jonny Grant
2023-02-19 20:43 ` Jonny Grant
2023-02-19 21:33 ` Jonny Grant
2023-02-20 10:26 ` Xi Ruoyao
2023-02-20 10:37 ` Jonathan Wakely
2023-02-20 10:54 ` Xi Ruoyao
2023-02-20 11:10 ` Gabriel Ravier
2023-02-20 11:18 ` Marc Glisse
2023-02-20 11:28 ` Segher Boessenkool
2023-02-20 12:00 ` Jonny Grant
2023-02-20 14:50 ` Gabriel Ravier
2023-02-20 11:44 ` Jonny Grant
2023-02-21 15:02 ` Jonny Grant
2023-02-20 11:38 ` Jonny Grant
2023-02-20 11:30 ` Jonny Grant
2023-02-20 12:59 ` Xi Ruoyao
2023-02-20 13:44 ` Jonathan Wakely
2023-02-20 19:21 ` Jonny Grant
2023-02-20 19:35 ` Jonathan Wakely
2023-02-20 19:39 ` Jonny Grant
2023-02-22 20:27 ` Jonny Grant
2023-02-21 15:04 ` Jonny Grant
2023-02-21 22:48 ` Jonny Grant
2023-03-04 15:00 ` Jonny Grant
2023-02-20 11:25 ` Jonny Grant
2023-03-12 22:10 ` Jonny Grant
2023-03-13 10:10 ` Jonathan Wakely [this message]
2023-03-13 19:55 ` Jonny Grant
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=CAH6eHdRkKWPCToJ+HoeiDJg2N2hCqz45c4Z1LUwG3ipaxNLBWw@mail.gmail.com \
--to=jwakely.gcc@gmail.com \
--cc=gcc-help@gcc.gnu.org \
--cc=jg@jguk.org \
--cc=xry111@xry111.site \
/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).