This is the mail archive of the gdb-prs@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

gdb/201: Revise definition of REGISTER_VIRTUAL_TYPE



>Number:         201
>Category:       gdb
>Synopsis:       Revise definition of REGISTER_VIRTUAL_TYPE
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    unassigned
>State:          open
>Class:          change-request
>Submitter-Id:   net
>Arrival-Date:   Thu Aug 23 22:48:00 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     ac131313@cygnus.com
>Release:        unknown-1.0
>Organization:
>Environment:

>Description:
REGISTER_VIRTUAL_TYPE doco should mention the target specific floatformat and intNN types and that the type selected should exactly match REGISTER_VIRTUAL_SIZE and REGISTER_RAW_SIZE.

REGISTER_CONVERTIBLE's definition should be refined so that it only applies to parameter conversions - when a parameter type doesn't match the registers virtual type - eg a float in an i387_ext register. value_of_register(), except for legacy code, shouldn't need to do any conversion.

The big user of the conversion stuff is findvar.c:value_from_register.  That also needs careful review.  I suspect that the current logic isn't sufficient to handle structures when passed using recent ABIs - a struct might be spread between several registers or even a register and memory.


>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]