lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 01 Feb 2024 09:16:59 +0100
From: "Fabio M. De Francesco" <fabio.maria.de.francesco@...ux.intel.com>
To: Peter Zijlstra <peterz@...radead.org>, dan.j.williams@...el.com,
 Dan Williams <dan.j.williams@...el.com>
Cc: linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...nel.org>,
 linux-cxl@...r.kernel.org, Ira Weiny <ira.weiny@...el.com>
Subject: Re: [RFC PATCH v3] cleanup: Add cond_guard() to conditional guards

On Thursday, 1 February 2024 02:12:12 CET Dan Williams wrote:
> Fabio M. De Francesco wrote:
> > I just noticed that this is not the final version. It misses a semicolon.
> > Please discard this v3. I'm sending v4.
> 
> Ok, but do please copy the aspect of scoped_conf_guard() to take a
> "_fail" statement argument. Passing a return code collector variable by
> reference just feels a bit too magical. I like the explicitness of
> passing the statement directly.

I had introduced a bug in my tests that made me see failures when there were 
not. Now I fixed it and tests don't fail.

I'm sending a new version that passes the return variable directly, not as a 
reference, similar but not equal to:

	cond_guard(..., rc, -EINTR, ...);

Actually, I'm doing this:

	cond_guard(..., rc, 0, -EINTR, ...);

I'm not passing 'rc = -EINTR' because I want to take into account the 
possibility that rc contains values different than 0 from previous assignments. 
I'm passing rc, so that the macro can assign either a success code or a 
failure error to this variable. Any value from previous assignments must be 
always overwritten: 

	#define cond_guard(_name, _ret, _scs, _err, args...) \
        	CLASS(_name, scope)(args); \
        	if (!__guard_ptr(_name)(&scope)) _ret = _err; \
        	else _ret = _scs;

I should have seen long ago that my tests were failing because of a missing 
'else' when passing a statement in 'cond_guard(..., rc = -EINTR, ...);'. It 
had nothing to do with how to pass 'rc'. Sorry for that confusion.

Fabio

Fabio 



Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ