This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
RE: [PATCH 06/10] vla: update type from newly created value
- From: "Agovic, Sanimir" <sanimir dot agovic at intel dot com>
- To: 'Pedro Alves' <palves at redhat dot com>
- Cc: "gdb-patches at sourceware dot org" <gdb-patches at sourceware dot org>
- Date: Wed, 20 Nov 2013 12:46:28 +0000
- Subject: RE: [PATCH 06/10] vla: update type from newly created value
- Authentication-results: sourceware.org; auth=none
- References: <1382366424-21010-1-git-send-email-sanimir dot agovic at intel dot com> <1382366424-21010-7-git-send-email-sanimir dot agovic at intel dot com> <528C8F1E dot 2080309 at redhat dot com>
> -----Original Message-----
> From: Pedro Alves [mailto:palves@redhat.com]
> Sent: Wednesday, November 20, 2013 11:30 AM
> To: Agovic, Sanimir
> Cc: gdb-patches@sourceware.org
> Subject: Re: [PATCH 06/10] vla: update type from newly created value
>
> On 10/21/2013 03:40 PM, Sanimir Agovic wrote:
> > Constructing a value based on a type and address might change the type
> > of the newly constructed value.
>
> OOC (and for the archives), why's that? Where does that occur?
>
Since dwarf3 certain attributes e.g upper/lower bound may be computed
dynamically. To support such attributes with the current gdb type-system
we require such types to be "normalized". This means types with dynamic
properties are converted to types with static properties.
To successful convert a type with dynamic properties into one with static
properties access to inferior memory is needed. Therefore we hooked into
the following value constructors
value_at/value_at_lazy/value_from_contents_and_address
as they require an inferior address in addition to a type to instantiate
a value. IIF the passed type has dynamic properties we resolve the bounds
and thus change the type.
Given the following statement:
struct value *val = value_at (my_vla_type, at_address);
Before this was always true:
TYPE_LENGTH (value_type (val)) == TYPE_LENGTH (my_vla_type)
This is not the case after applying this (vla-c00) patch series. Type
normalization is done in the mentioned value constructors and might
change the value type, therefore we need to re-fetch the type from
the constructed value.
I hope this makes sense.
> > Thus re-fetch type via value_type to ensure
> > we have the correct type at hand.
>
> --
> Pedro Alves
Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen, Deutschland
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk
Registergericht: Muenchen HRB 47456
Ust.-IdNr./VAT Registration No.: DE129385895
Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052