At <https://sourceware.org/ml/gdb-patches/2017-12/msg00298.html>, Joel
wrote:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Consider the following code which first declares a tagged type (the
equivalent of a class in Ada), and then a procedure which takes a
pointer (access) to this type's 'Class.
package Pck is
type Top_T is tagged record
N : Integer := 1;
end record;
procedure Inspect (Obj: access Top_T'Class);
end Pck;
Putting a breakpoint in that procedure and then running to it triggers
an internal error:
What's special about this subprogram is that it takes an access to
what we call a 'Class type, and for implementation reasons, the
compiler adds an extra argument named "objL". If you are curious why,
it allows the compiler for perform dynamic accessibility checks that
are mandated by the language.
If we look at the location where we get the internal error (in
stack.c), we find that we are looping over the symbol of each
parameter, and for each parameter, we do:
/* We have to look up the symbol because arguments can have
two entries (one a parameter, one a local) and the one we
want is the local, which lookup_symbol will find for us.
[...]
nsym = lookup_symbol (SYMBOL_LINKAGE_NAME (sym),
b, VAR_DOMAIN, NULL).symbol;
gdb_assert (nsym != NULL);
The lookup_symbol goes through the lookup structure, which means the
symbol's linkage name ("objL") gets transformed into a
lookup_name_info object (in block_lookup_symbol), before it gets fed
to the block symbol dictionary iterators. This, in turn, triggers the
symbol matching by comparing the "lookup" name which, for Ada, means
among other things, lowercasing the given name to "objl". It is this
transformation that causes the lookup find no matches, and therefore
trip this assertion.
Going back to the "offending" call to lookup_symbol in stack.c, what
we are trying to do, here, is do a lookup by linkage name. So, I
think what we mean to be doing is a completely literal symbol lookup,
so maybe not even strcmp_iw, but actually just plain strcmp???
In the past, in practice, you could get that effect by doing a lookup
using the C language. But that doesn't work, because we still end up
somehow using Ada's lookup_name routine which transforms "objL".
So, ideally, as I hinted before, I think what we need is a way to
perform a literal lookup so that searches by linkage names like the
above can be performed.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This commit fixes the problem by implementing something similar to
Joel's literal idea, but with some important differences.
I considered adding a symbol_name_match_type::LINKAGE and supporting
searching by linkage name for any language, but the problem with that
is that the dictionaries only work with SYMBOL_SEARCH_NAME, because
that's what is used for hashing. We'd need separate dictionaries for
hashed linkage names.
So with the current symbol tables infrastructure, it's not literal
linkage names that we want to pass down, but instead literal _search_
names (SYMBOL_SEARCH_NAME, etc.).
However, psymbols have no overload/function parameter info in C++, so
a straight strcmp doesn't work properly for C++ name matching.
So what we do is be a little less aggressive then and add a new
symbol_name_match_type::SEARCH_SYMBOL instead that takes as input a
non-user-input search symbol, and then we skip any decoding/demangling
steps and make:
- Ada treat that as a verbatim match,
- other languages treat it as symbol_name_match_type::FULL.
This also fixes the new '"maint check-psymtabs" for Ada' testcase for
me (gdb.ada/maint_with_ada.exp). I've not removed the kfail yet
because Joel still sees that testcase failing with this patch.
That'll be fixed in follow up patches.
gdb/ChangeLog:
2018-01-05 Pedro Alves <palves@redhat.com>
PR gdb/22670
* ada-lang.c (literal_symbol_name_matcher): New function.
(ada_get_symbol_name_matcher): Use it for
symbol_name_match_type::SEARCH_NAME.
* block.c (block_lookup_symbol): New parameter 'match_type'. Pass
it down instead of assuming symbol_name_match_type::FULL.
* block.h (block_lookup_symbol): New parameter 'match_type'.
* c-valprint.c (print_unpacked_pointer): Use
lookup_symbol_search_name instead of lookup_symbol.
* compile/compile-object-load.c (get_out_value_type): Pass down
symbol_name_match_type::SEARCH_NAME.
* cp-namespace.c (cp_basic_lookup_symbol): Pass down
symbol_name_match_type::FULL.
* cp-support.c (cp_get_symbol_name_matcher): Handle
symbol_name_match_type::SEARCH_NAME.
* infrun.c (insert_exception_resume_breakpoint): Use
lookup_symbol_search_name.
* p-valprint.c (pascal_val_print): Use lookup_symbol_search_name.
* psymtab.c (maintenance_check_psymtabs): Use
symbol_name_match_type::SEARCH_NAME and SYMBOL_SEARCH_NAME.
* stack.c (print_frame_args): Use lookup_symbol_search_name and
SYMBOL_SEARCH_NAME.
* symtab.c (lookup_local_symbol): Don't demangle the lookup name
if symbol_name_match_type::SEARCH_NAME.
(lookup_symbol_in_language): Pass down
symbol_name_match_type::FULL.
(lookup_symbol_search_name): New.
(lookup_language_this): Pass down
symbol_name_match_type::SEARCH_NAME.
(lookup_symbol_aux, lookup_local_symbol): New parameter
'match_type'. Pass it down.
* symtab.h (symbol_name_match_type::SEARCH_NAME): New enumerator.
(lookup_symbol_search_name): New declaration.
(lookup_symbol_in_block): New 'match_type' parameter.
gdb/testsuite/ChangeLog:
2018-01-05 Joel Brobecker <brobecker@adacore.com>
PR gdb/22670
* gdb.ada/access_tagged_param.exp: New file.
* gdb.ada/access_tagged_param/foo.adb: New file.
The comments about mixed case in the testcase are actually a red
herring. The problem here is that we'd get to
ada_lookup_encoded_symbol with "my_table", which wraps the looked up
name in "<>"s to force a verbatim match, and that in turn disables
wild matching.
Fix this by swapping around the internals of ada_lookup_encoded_symbol
and ada_lookup_symbol, thus avoiding the encoding and
verbatim-wrapping in the ada_lookup_symbol case, the case that starts
with a user-provided lookup name.
Ada encoding is still done of course, in the ada_lookup_name_info
ctor. This could be also seen as avoiding the double-encoding problem
in a different way.
gdb/ChangeLog:
2018-01-05 Pedro Alves <palves@redhat.com>
PR gdb/22670
* ada-lang.c (ada_lookup_encoded_symbol): Reimplement in terms of
ada_lookup_symbol.
(ada_lookup_symbol): Reimplement in terms of
ada_lookup_symbol_list, bits factored out from
ada_lookup_encoded_symbol.
gdb/testsuite/ChangeLog:
2018-01-05 Pedro Alves <palves@redhat.com>
PR gdb/22670
* gdb.ada/info_addr_mixed_case.exp: Remove kfail. Extend test to
exercise lower case too, and to exercise both full matching and
wild matching.
Joel Brobecker [Fri, 5 Jan 2018 09:29:54 +0000 (04:29 -0500)]
(Ada) problem printing renaming which references a subprogram parameter
Consider the following code, which creates a local variable B
which is a renaming whose expression references a subprogram
parameter:
procedure Flip (Bits : in out Bits_Type; I : Natural) is
begin
declare
B : Boolean renames Bits (I);
begin
B := not B; -- BREAK
end;
end Flip;
Trying to print the value of B when at the "BREAK" line currently
does not work:
(gdb) p b
Could not find i
What happens is the following: For the renaming, GNAT generates
a variable whose name is encoded as follow:
b___XR_bits___XEXSi
GDB properly detects that variable, determines that, to compute
the variable's value, we start from the symbol "Bits", which
we then have to subscript (XS) using 'i' as the index. The error
occurs while trying to find 'i'.
This is because we forgot to pass the block in the call to
ada_lookup_encoded_symbol, which this patch fixes.
gdb/ChangeLog:
* ada-exp.y (write_object_renaming): When subscripting an array
using a symbol as the index, pass the block in call to
ada_lookup_encoded_symbol when looking that symbol up.
gdb/testsuite/ChangeLog:
* gdb.ada/rename_subscript_param: New testcase.
Tested on x86_64-linux.
Note: This requires the following GCC patch:
| 2017-04-25 Pierre-Marie de Rodat <derodat@adacore.com>
|
| * exp_dbug.adb: In Debug_Renaming_Declaration,
| when dealing with indexed component, accept to produce a renaming
| symbol when the index is an IN parameter or when it is a name
| defined in an outer scope.
Jerome Guitton [Fri, 5 Jan 2018 08:03:59 +0000 (03:03 -0500)]
(Ada) Fix Length attribute on array access
Consider the following variable "Indexed_By_Enum", declared as
an access to an array whose index type is an enumerated type
whose underlying values have "gaps":
type Enum_With_Gaps is (LIT0, LIT1, LIT2, LIT3, LIT4);
for Enum_With_Gaps use (LIT0 => 3,
LIT1 => 5,
LIT2 => 8,
LIT3 => 13,
LIT4 => 21);
for Enum_With_Gaps'size use 16;
type MyWord is range 0 .. 16#FFFF# ;
for MyWord'Size use 16;
type AR is array (Enum_With_Gaps range <>) of MyWord;
type AR_Access is access AR;
Trying to print the length (number of elements) of this array using
the 'Length attribute does not work:
(gdb) print indexed_by_enum'length
'POS only defined on discrete types
The problem occurs while trying to get the array's index type.
It was using TYPE_INDEX_TYPE for that. It does not work for Ada arrays
in general; use ada_index_type instead.
gdb/ChangeLog:
* ada-lang.c (ada_array_length): Use ada_index_type instead of
TYPE_INDEX_TYPE.
Joel Brobecker [Fri, 5 Jan 2018 07:37:38 +0000 (02:37 -0500)]
memory error printing component of record from convenience variable
Consider the following situation Ada code:
type Kind_T is (One, Two, Three);
type Time_Set_T is array (Kind_T) of Integer;
type T is record
Started : Time_Set_T;
end record;
Null_T : constant T := (Started => (others => 0));
My_Item : Pck.T := Pck.Null_T;
Trying to print the value of My_Item.Started is no problem:
(gdb) p item.started
$1 = (0, 0, 0)
However, if you save My_Item into a convenience variable first,
and then try to print a component of that record, you get
an unexpected memory error, instead of getting the same result.
For instance:
(gdb) set variable $item := item
(gdb) p $item.started
Cannot access memory at address 0x0
The issue occurs when, after we extracted the component from
the convenience variable, we next try to "fix" it (which is
ada-lang speak for resolving the type into a static type).
This is done in ada_to_fixed_value, which delegates to
ada_to_fixed_value_create via:
val = ada_to_fixed_value_create (value_type (val),
value_address (val), val);
And looking at ada_to_fixed_value_create, we see that:
struct type *type = ada_to_fixed_type (type0, 0, address, NULL, 1);
The part that interests us, in this case, is the "else" branch,
where we obviously make the implicit assumption that our object
has an address, which is not true, in this case, because we are
actually dealing with a convenience variable.
This patch plugs that hole by adding special handing for situations
where val does not live in memory. In that case, we just create
a not_lval value using val's contents.
gdb/ChangeLog:
* ada-lang.c (ada_to_fixed_value_create): Add handling of
the case where VALUE_LVAL (val0) is not lval_memory.
In this case, the index of the first element in this array is also
the first element of the index type (character type). Similar to what
we do we enumeration types, we do not need to print the index of the
first element when printing the array.
This patch fixes this issue and changes the output as follow:
include/
* opcode/riscv-opc.h (CSR_SBADADDR): Rename to CSR_STVAL. Rename
DECLARE_CSR entry. Add alias to map sbadaddr to CSR_STVAL.
(CSR_MBADADDR): Rename to CSR_MTVAL. Rename DECLARE_CSR entry.
Add alias to map mbadaddr to CSR_MTVAL.
Joel Brobecker [Thu, 4 Jan 2018 09:40:17 +0000 (04:40 -0500)]
gdb.ada/maint_with_ada.exp: New testcase
This commit adds a new testcase testing the "check-psymtabs" and
"check-symtabs" maintenance commands.
The "maintenance check-psymtabs" commands is currently known to
produce some errors. While the situation was admetedly made worse
by the following patch...
commit b5ec771e60c1a0863e51eb491c85c674097e9e13
Date: Wed Nov 8 14:22:32 2017 +0000
Subject: Introduce lookup_name_info and generalize Ada's FULL/WILD name matching
... hence the reference to PR gdb/22670, the command was already
spotting one inconsistency prior to it:
(gdb) maintenance check-psymtabs
Global symbol `interfaces__cS' only found in /[...]/b~var_arr_typedef.adb psymtab
For now, the "check-psymtab" test is KFAIL-ed.
gdb/testsuite/ChangeLog:
PR gdb/22670
* gdb.ada/maint_with_ada: New testcase.
Joel Brobecker [Thu, 4 Jan 2018 08:10:34 +0000 (03:10 -0500)]
Add new gdb.ada/bp_c_mixed_case testcase for PR gdb/22670
This patch adds a new testcase to demonstrate a regression introduced by:
commit b5ec771e60c1a0863e51eb491c85c674097e9e13
Date: Wed Nov 8 14:22:32 2017 +0000
Subject: Introduce lookup_name_info and generalize Ada's FULL/WILD name matching
The purpose of the testcase is to verify that a user can insert
a breakpoint on a C function while debugging Ada, even if the name
of the function includes uppercase letters, requiring us to use
Ada's "<...>" notation to tell the GDB that the symbol name should
be looked up verbatim.
As of the commit above, GDB is no longer finding the function:
(gdb) break <MixedCaseFunc>
Function "<MixedCaseFunc>" not defined.
Make breakpoint pending on future shared library load? (y or [n])
Before the patch, the breakpoint was inserted without problem.
gdb/testsuite/ChangeLog:
PR gdb/22670
* gdb.ada/bp_c_mixed_case: New testcase.
Tested on x86_64-linux; generates a KPASS before the regression
was introduced, and now generates a KFAIL.
Joel Brobecker [Thu, 4 Jan 2018 07:39:58 +0000 (02:39 -0500)]
Add "complete break ada" test to gdb.ada/complete.exp
This patch adds a new test to demonstrate a regression introduced by:
commit b5ec771e60c1a0863e51eb491c85c674097e9e13
Date: Wed Nov 8 14:22:32 2017 +0000
Subject: Introduce lookup_name_info and generalize Ada's FULL/WILD name matching
The original purpose of the new test is to exercise the "complete"
command with an expression for which a large number of matches are
returned and to verify that each match returned is a plausible match.
In this particular case, the commit above causes GDB to generate
additional matches which should in fact not appear in the list
(internally generated symbols, or symbols that should be enclosed
between "<...>"). These extraneous entries are easy to spot, because
they have uppercase characters, such as:
Tested on x86_64-linux with GDB built before and after the patch
that caused the regression (b5ec771e60c1a0863e51eb491c85c674097e9e13).
The test passes before the regression, and generates a KFAIL after.
Joel Brobecker [Thu, 4 Jan 2018 04:28:14 +0000 (23:28 -0500)]
Add gdb.ada/info_addr_mixed_case new testcase
This patch adds a new testcase to demonstrate a regression introduced by:
commit b5ec771e60c1a0863e51eb491c85c674097e9e13
Date: Wed Nov 8 14:22:32 2017 +0000
Subject: Introduce lookup_name_info and generalize Ada's FULL/WILD name matching
The purpose of the testcase is to verify that a user can use any
casing for an Ada symbol name passed to the "info address" command.
After the patch above was applied, GDB was no longer able to find
the symbol:
(gdb) info address My_Table
No symbol "My_Table" in current context.
gdb/testsuite/ChangeLog:
PR gdb/22670
* gdb.ada/info_addr_mixed_case: New testcase.
Tested on x86_64-linux, both before and after the patch.
Nick Clifton [Wed, 3 Jan 2018 10:28:33 +0000 (10:28 +0000)]
Add support for v3 binary annotation notes.
I am checking in the attached patch which updates the binutils
support for version 3 binary annotation notes. (Version 3 adds
an end address to the ranges covered by the notes, so that it
is possible to detect gaps in the coverage).
This patch also stops the note merging feature of objcopy from
executing if the notes have relocations against them. This makes the
code simpler, and prevents the problems with architectures which have
unusual relocation management issues.
* objcopy.c (objcopy_internal_note): New structure.
(gap_exists): New function.
(is_open_note): New function.
(is_func_note): New function.
(is_64bit): New function.
(merge_gnu_build_notes): Handle v3 notes. Do not merge
if there are relocations against the notes.
* readelf.c (get_note_type): Use short names for build attribute
notes.
(print_symbol_for_build_attribute): Rename to
get_symbol_for_build_attribute. Returns the found symbol rather
than printing it.
(print_gnu_build_attribute_description): Maintain address ranges
for function notes as well as global notes. Handle v3 notes.
(print_gnu_build_attribute_name): Use more space for printing the
name in wide mode.
* testsuite/binutils-all/note-2-32.s: Use .dc.l instead of .word.
Eliminate symbol references in order to remove the need for
relocations.
* testsuite/binutils-all/note-2-64.s: Likewise.
* testsuite/binutils-all/note-3-32.s: Add a size to the note_1
symbol.
* testsuite/binutils-all/note-3-64.s: Likewise.
* testsuite/binutils-all/mips/mips-note-2r-n32.d: Update expected
output.
* testsuite/binutils-all/mips/mips-note-2r-n64.d: Likewise.
* testsuite/binutils-all/mips/mips-note-2r.d: Likewise.
* testsuite/binutils-all/note-2-32.d: Likewise.
* testsuite/binutils-all/note-2-64.d: Likewise.
* testsuite/binutils-all/note-3-32.d: Likewise.
* testsuite/binutils-all/note-3-64.d: Likewise.
* testsuite/binutils-all/note-4-64.s: New test. Checks v3 notes.
* testsuite/binutils-all/note-4-32.s: New test.
* testsuite/binutils-all/note-4-64.d: New test result file.
* testsuite/binutils-all/note-4-32.d: New test result file.
Xavier Roirand [Wed, 22 Nov 2017 09:40:39 +0000 (10:40 +0100)]
(Ada) New command to stop at start of exception handler.
When using gdb for debugging Ada source code, there are several catchpoint
types you can define in order to stop upon certain conditions. Let's
use this small example:
procedure Foo is
begin
begin
raise Constraint_Error;
exception
when Program_Error =>
null;
when Constraint_Error =>
null;
when others =>
null;
end;
end Foo;
One can stop when the exception is being raised by using the exception
catchpoint like below:
(gdb) catch exception
Catchpoint 1: all Ada exceptions
(gdb)
In that case, when running Foo, gdb will stop at the line where the exception
was raised:
begin
>>> raise Constraint_Error;
exception
This patch introduces new type of catchpoint, when the user wants to stop
at the location of the exception handling.
Imagine we want to stop on any exception handled by the program, we can do:
(gdb) catch handlers
Catchpoint 1: all Ada exceptions handlers
(gdb) r
Starting program: /tmp/foo
By doing so, when running Foo, gdb will stop here:
Catchpoint 1, exception at 0x000000000040255a in foo () at foo.adb:25
25 when Constraint_Error =>
(gdb)
It is also possible to stop when the Constraint_Error exception is being
handled in this program. With this patch, we can use:
Joel Brobecker [Wed, 3 Jan 2018 03:50:38 +0000 (07:50 +0400)]
Unbreak spurious fails in gdb.base/step-line.exp
My copyright year update in the copyright headers of the step-line
testcase caused it to start failing, due to shift in line numbers.
FAIL: gdb.base/step-line.exp: continue to f1
FAIL: gdb.base/step-line.exp: next to dummy 2
FAIL: gdb.base/step-line.exp: next over dummy 2
FAIL: gdb.base/step-line.exp: step into f2
FAIL: gdb.base/step-line.exp: next over dummy 4
FAIL: gdb.base/step-line.exp: next to dummy 5
FAIL: gdb.base/step-line.exp: next to dummy 6
FAIL: gdb.base/step-line.exp: next over dummy 6
FAIL: gdb.base/step-line.exp: next to dummy 7
FAIL: gdb.base/step-line.exp: next to dummy 8
FAIL: gdb.base/step-line.exp: next over dummy 8
FAIL: gdb.base/step-line.exp: next to dummy 9
FAIL: gdb.base/step-line.exp: next to dummy 10
FAIL: gdb.base/step-line.exp: next over dummy 10
This was caused by the copyright.py script after I changed it to
also update this testcase, which made the following transformation:
- Copyright (C) 2001-2017
- Free Software Foundation, Inc.
+ Copyright (C) 2001-2018 Free Software Foundation, Inc.
We can easily unbreak the testcase without having to go back to
manual edits each year by adding an extra line in the header.
gdb/testsuite/ChangeLog:
* gdb.base/step-line.c: Add extra empty line in copyright header.
* gdb.base/step-line.inp: Likewise.
Other than the variables in tc-i386.c using them, OPERAND_TYPE_REGYMM
and OPERAND_TYPE_REGZMM they aren't entirely unused. No need to update
i386-init.h though, as it mistakenly wasn't updated by the original
commit.
Joel Brobecker [Tue, 2 Jan 2018 03:53:55 +0000 (22:53 -0500)]
[gdb/Ada] slices of arrays with dynamic strides
Consider the following Ada code:
procedure Nested (L, U : Integer) is
subtype Small_Type is Integer range L .. U;
type Record_Type (I : Small_Type := L) is record
S : String (1 .. I);
end record;
type Array_Type is array (Integer range <>) of Record_Type;
A1 : Array_Type :=
(1 => (I => 0, S => <>),
2 => (I => 1, S => "A"),
3 => (I => 2, S => "AB"));
procedure Discard (R : Record_Type) is
begin
null;
end Discard;
begin
Discard (A1 (1)); -- STOP
end;
Trying to print a slice of that array currently yields:
(gdb) p a1(1..3)
$1 = ((i => 0, s => ""), (i => 0, s => ""), (i => 0, s => ""))
We expected instead:
(gdb) p a1(1..3)
$1 = ((i => 0, s => ""), (i => 1, s => "A"), (i => 2, s => "AB"))
This is because the functions we use in ada-lang.c to create the type
of the array slice (ada_value_slice and ada_value_slice_from_ptr) was
not taking into account the stride of the array. This patch fixes this.
gdb/ChangeLog:
* ada-lang.c (ada_value_slice_from_ptr): Take array stride into
account when creating the array type of the slice.
(ada_value_slice): Likewise.
gdb/testsuite/ChangeLog:
* gdb.ada/dyn_stride.exp: Add slice test.
Note that, with the current use of ada_value_slice, the enhancement
to handle dynamic array strides seems unnecessary, because I do not
see how an array with a dynamic stride can be referenced by either
by reference or pointer. Since references are coerced to array pointers,
in both cases, the slice is performed by ada_value_slice_from_ptr.
But ada_value_slice is enhanced nonetheless, in the spirit of making
the code more robust, in case we missed something, and also as similar
as possible with its from_ptr counterpart.
Joel Brobecker [Tue, 2 Jan 2018 03:47:18 +0000 (22:47 -0500)]
Add support for dynamic DW_AT_byte_stride.
This patch adds support for DW_AT_byte_stride, using Ada as one
example of where this would be useful. However, the implementation
is language-agnostic.
Consider the following Ada code:
procedure Nested (L, U : Integer) is
subtype Small_Type is Integer range L .. U;
type Record_Type (I : Small_Type := L) is record
S : String (1 .. I);
end record;
type Array_Type is array (Integer range <>) of Record_Type;
A1 : Array_Type :=
(1 => (I => 0, S => <>),
2 => (I => 1, S => "A"),
3 => (I => 2, S => "AB"));
procedure Discard (R : Record_Type) is
begin
null;
end Discard;
begin
Discard (A1 (1)); -- STOP
end;
It defines an array A1 of Record_Type, which is a variant record
type whose maximum size actually depends on the value of the
parameters passed when calling Nested. As a result, the stride
of the array A1 cannot be known statically, which leads the compiler
to generate a dynamic DW_AT_byte_stride attribute for our type.
Here is what the debugging info looks like with GNAT:
.uleb128 0x10 # (DIE (0x14e) DW_TAG_array_type)
.long .LASF17 # DW_AT_name: "foo__nested__T18b"
.long 0x141 # DW_AT_byte_stride
.long 0xdc # DW_AT_type
.uleb128 0x11 # (DIE (0x15f) DW_TAG_subrange_type)
.long 0x166 # DW_AT_type
.byte 0x3 # DW_AT_upper_bound
.byte 0 # end of children of DIE 0x14e
There DW_AT_byte_stride is a reference to a local (internal)
variable:
This patch enhances GDB to handle this dynamic byte stride attribute
by first adding a new dynamic_prop_node_kind (DYN_PROP_BYTE_STRIDE)
to store the array dynamic stride info (when dynamic). It then enhances
the dynamic type resolver to handle this dynamic property.
Before applying this patch, trying to print the value of some of
A1's elements after having stopped at the "STOP" comment does not
work. For instance:
(gdb) p a1(2)
Cannot access memory at address 0x80000268dec0
With this patch applied, GDB now prints the value of all 3 elements
correctly:
(gdb) print A1(1)
$1 = (i => 0, s => "")
(gdb) print A1(2)
$2 = (i => 1, s => "A")
(gdb) print A1(3)
$3 = (i => 2, s => "AB")
gdb/ChangeLog:
* gdbtypes.h (enum dynamic_prop_node_kind) <DYN_PROP_BYTE_STRIDE>:
New enum value.
(create_array_type_with_stride): Add byte_stride_prop parameter.
* gdbtypes.c (create_array_type_with_stride) <byte_stride_prop>:
New parameter. Update all callers in this file.
(array_type_has_dynamic_stride): New function.
(is_dynamic_type_internal, resolve_dynamic_array): Add handling
of arrays with dynamic byte strides.
* dwarf2read.c (read_array_type): Add support for dynamic
DW_AT_byte_stride attributes.
Joel Brobecker [Tue, 2 Jan 2018 03:41:36 +0000 (22:41 -0500)]
treat Ada DW_TAG_unspecified_type DIEs as stub types
Consider the gdb.ada/taft_type.exp testcase, which exercises
the situation where a variable is defined using a type which
is a pointer to an incomplete type, with the actual type
definition being provided by another unit. Up to now, the
strategy used by GNAT when generating the DWARF debugging info
was to produce a incomplete DW_TAG_enumeration_type DIE with
a DW_AT_declaration flag attached to it:
However, a more standard way for the compiler to describe
this kind of type is to use the DW_TAG_unspecified_type tag.
When the compiler is enhanced to do so, we'll need to treat
such types as stubs -- we only do so with types from Ada units,
however, as the meaning of this TAG is intentionally left
permissive and language-specific by the DWARF standard.
Without this patch, running the testcase above with an enhanced
compiler now yields:
(gdb) print w.e.all
Attempt to dereference a generic pointer.
FAIL: gdb.ada/taft_type.exp: print w.e.all
gdb/ChangeLog:
* dwarf2read.c (read_unspecified_type): Treat
DW_TAG_enumeration_type DIEs from Ada units as stubs.
Tested on x86_64-linux, fixes the FAIL in gdb.ada/taft_type.exp above.
Joel Brobecker [Mon, 1 Jan 2018 04:18:16 +0000 (08:18 +0400)]
gdb/copyright.py: Remove testsuite/gdb.base/step-line.{c,inp} special handling
In the past, these files needed to be handled by hand, because
the testcase was sensitive to the length of the header, which was
potentially changing when new copyright years were added to
the copyright header. Now that we simply maintain and update
a range, the length of the copyright header should not change
as a consequence of the update performed by this script, so
special handling of those files is no longer necessary.
gdb/ChangeLog:
* copyright.py (BY_HAND): Remove gdb/testsuite/gdb.base/step-line.inp
and gdb/testsuite/gdb.base/step-line.c.
Joel Brobecker [Mon, 1 Jan 2018 04:30:28 +0000 (08:30 +0400)]
gdb/copyright.py: Do not forget to remind about MULTIPLE_COPYRIGHT_HEADERS
There is a small logical error in the part of the script that
dumps the list of files in BY_HAND + MULTIPLE_COPYRIGHT_HEADERS
but only checkis the contents of BY_HAND. The issue becomes
apparent as soon as BY_HAND is empty. Prevent this from happening
by treating the two lists separately, as this allows us to provide
a more informative message in the case of MULTIPLE_COPYRIGHT_HEADERS.
gdb/ChangeLog:
* copyright.py (main): Dump the contents of
MULTIPLE_COPYRIGHT_HEADERS (separately) from BY_HAND,
even if BY_HAND is empty.
The code here wants to find address of an element, and often this
element is one past the end of std::vector. Dereferencing that element
leads to undefined behavior, so it's better to simply use pointer
arithmetic instead of taking address of invalid dereference.
gdb/ChangeLog:
* psymtab.c (recursively_search_psymtabs): Use pointer arithmetic
instead of dereferencing std::vector past the end.
made some parts of dwarf2read.c ignore warnings about switch using enums
of different kinds. What I did not realize was that older Clang
versions (prior to 6) did not have that warning, and therefore give this
error:
/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:24187:7: error: unknown warning group '-Wenum-compare-switch', ignored [-Werror,-Wunknown-pragmas]
DIAGNOSTIC_IGNORE_SWITCH_DIFFERENT_ENUM_TYPES
^
/home/emaisin/src/binutils-gdb/gdb/common/diagnostics.h:42:3: note: expanded from macro 'DIAGNOSTIC_IGNORE_SWITCH_DIFFERENT_ENUM_TYPES'
DIAGNOSTIC_IGNORE ("-Wenum-compare-switch")
^
/home/emaisin/src/binutils-gdb/gdb/common/diagnostics.h:27:3: note: expanded from macro 'DIAGNOSTIC_IGNORE'
_Pragma (STRINGIFY (GCC diagnostic ignored option))
^
<scratch space>:10:25: note: expanded from here
GCC diagnostic ignored "-Wenum-compare-switch"
^
Clang has a way to test if it knows about a particular warning. This
patch uses that feature to only define
DIAGNOSTIC_IGNORE_SWITCH_DIFFERENT_ENUM_TYPES to something if the
warning is recognized by the Clang version being used. I tested
building dwarf2read.c with clang 4, 5, 6, as well as gcc.
gdb/ChangeLog:
* common/diagnostics.h
(DIAGNOSTIC_IGNORE_SWITCH_DIFFERENT_ENUM_TYPES): Only
define if the compiler knows about -Wenum-compare-switch.
Simon Marchi [Sun, 31 Dec 2017 02:24:29 +0000 (21:24 -0500)]
dwarf2read: Silence -Wenum-compare-switch warning
Compiling with Clang 6 gives:
/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:24385:14: error: comparison of two values with different enumeration types in switch statement ('enum dwarf_macro_record_type' and 'dwarf_macinfo_record_type') [-Werror,-Wenum-compare-switch]
case DW_MACINFO_vendor_ext:
^~~~~~~~~~~~~~~~~~~~~
/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:24561:7: error: comparison of two values with different enumeration types in switch statement ('enum dwarf_macro_record_type' and 'dwarf_macinfo_record_type') [-Werror,-Wenum-compare-switch]
case DW_MACINFO_vendor_ext:
^~~~~~~~~~~~~~~~~~~~~
This code uses the two enum types on purpose, because it handles both
.debug_macro and .debug_macinfo sections. Add some pragmas to disable
the warning in these specific cases.
Simon Marchi [Sat, 30 Dec 2017 05:10:05 +0000 (00:10 -0500)]
Make mapped_debug_names and mapped_index final
When compiling with Clang 6, I see these warnings:
/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:25421:5: error: destructor called on non-final 'mapped_index' that has virtual functions but non-virtual destructor [-Werror,-Wdelete-non-virtual-dtor]
data->index_table->~mapped_index ();
^
In file included from /home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:31:
In file included from /home/emaisin/src/binutils-gdb/gdb/defs.h:28:
In file included from /home/emaisin/src/binutils-gdb/gdb/common/common-defs.h:92:
In file included from /home/emaisin/src/binutils-gdb/gdb/common/gdb_unique_ptr.h:23:
In file included from /usr/bin/../lib/gcc/x86_64-linux-gnu/5.4.0/../../../../include/c++/5.4.0/memory:81:
/usr/bin/../lib/gcc/x86_64-linux-gnu/5.4.0/../../../../include/c++/5.4.0/bits/unique_ptr.h:76:2: error: delete called on non-final 'mapped_debug_names' that has virtual functions but non-virtual destructor [-Werror,-Wdelete-non-virtual-dtor]
delete __ptr;
^
/usr/bin/../lib/gcc/x86_64-linux-gnu/5.4.0/../../../../include/c++/5.4.0/bits/unique_ptr.h:236:4: note: in instantiation of member function 'std::default_delete<mapped_debug_names>::operator()' requested here
get_deleter()(__ptr);
^
/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:2374:21: note: in instantiation of member function 'std::unique_ptr<mapped_debug_names, std::default_delete<mapped_debug_names> >::~unique_ptr' requested here
dwarf2_per_objfile::dwarf2_per_objfile (struct objfile *objfile_,
^
This patch silences these warnings by making the classes final.
gdb/ChangeLog:
* dwarf2read.c (struct mapped_debug_names): Make final.
(struct mapped_index): Make final.
Simon Marchi [Sat, 30 Dec 2017 04:32:13 +0000 (23:32 -0500)]
Ignore warning about using different types of enums in switch
When compiling with clang 6, I see a bunch of warnings like this:
/home/emaisin/src/binutils-gdb/gdb/amd64-linux-tdep.c:1427:8: error: comparison of two values with different enumeration types in switch statement ('enum amd64_syscall' and 'amd
64_x32_syscall') [-Werror,-Wenum-compare-switch]
case amd64_x32_sys_move_pages:
^~~~~~~~~~~~~~~~~~~~~~~~
In this switch, we indeed use enumerators of both types
amd64_x32_syscall and amd64_syscall. This is done on purpose, and the
enum values are chosen so that they are complementary.
I think it's still a useful warning, so I chose to ignore just that
particular case.
gdb/ChangeLog:
* common/diagnostics.h
(DIAGNOSTIC_IGNORE_SWITCH_DIFFERENT_ENUM_TYPES): New macro.
* amd64-linux-tdep.c (amd64_canonicalize_syscall): Use it.
Simon Marchi [Sat, 30 Dec 2017 02:56:58 +0000 (21:56 -0500)]
Remove unnecessary call to get_thread_db_info
In thread_db_detach, we call get_thread_db_info to first check if there
exists a thread_db_info entry for the pid to detach. If there is, then
we call delete_thread_db_info. It's unnecessary to call
get_thread_db_info in the first place, since delete_thread_db_info
handles the case where no thread_db_info entry exist for the given pid.
gdb/ChangeLog:
* linux-thread-db.c (thread_db_detach): Remove call to
delete_thread_db_info.
Stafford Horne [Wed, 27 Dec 2017 14:36:31 +0000 (23:36 +0900)]
tdesc: handle arbitrary strings in tdesc_register_in_reggroup_p
tdesc_register_in_reggroup_p in now able to handle arbitrary
groups. This is useful when groups are created while the
target descriptor file is received from the remote.
This can be the case of a soft core target processor where
registers/groups can change.
* target-descriptions.c (tdesc_register_in_reggroup_p): Support
arbitrary strings.
(tdesc_use_registers): Add logic to register new reggroups.
(tdesc_reg::group): Update comment to indicate we allow
arbitrary strings.
* NEWS (Changes since GDB 8.0): Announce that GDB supports
arbitrary reggroups.
gdb/testsuite/ChangeLog:
yyyy-mm-dd Stafford Horne <shorne@gmail.com>
* gdb.xml/extra-regs.xml: Add example foo reggroup.
* gdb.xml/tdesc-regs.exp: Add test to check for foo reggroup.
gdb/doc/ChangeLog:
yyyy-mm-dd Stafford Horne <shorne@gmail.com>
* gdb.texinfo (Target Description Format): Explain that arbitrary
strings are now allowed for register groups.
Stafford Horne [Wed, 27 Dec 2017 14:36:31 +0000 (23:36 +0900)]
reggroups: Add reggroup_gdbarch_new, reggroup_find for dynamic reggroups
Traditionally reggroups have been created via reggroup_new() during
initialization code and never freed. Now, if we want to initialize
reggroups dynamically (i.e. in target description) we should be able to
free them. Create this function reggroup_gdbarch_new() which will
allocate the reggroup memory onto the passed gdbarch obstack.
Also creating reggroup_find() as a utility to find a gdbarch registered
reggroup object by name.
gdb/ChangeLog:
yyyy-mm-dd Stafford Horne <shorne@gmail.com>
* reggroups.c (reggroup_gdbarch_new): New function.
(reggroup_find): New function.
* reggroups.h (reggroup_gdbarch_new): New function.
(reggroup_find): New function.
Stafford Horne [Wed, 27 Dec 2017 14:36:30 +0000 (23:36 +0900)]
reggroups: Convert reggroups from post_init to pre_init
Currently the reggroups gdbarch_data cannot be manipulated until after
the gdbarch is completely initialized. This is usually done when the
object init depends on architecture specific fields. In the case of
reggroups it only depends on the obstack being available.
Coverting this to pre_init allows using reggroups during gdbarch
initialization. This is needed to allow registering arbitrary reggroups
during gdbarch initializations.
gdb/ChangeLog:
yyyy-mm-dd Stafford Horne <shorne@gmail.com>
* reggroups.c (reggroups_init): Change to depend only on
obstack rather than gdbarch.
(reggroup_add): Remove logic for forcing premature init.
(_initialize_reggroup): Set `reggroups_data` with
gdbarch_data_register_pre_init() rather than
gdbarch_data_register_post_init().
Simon Marchi [Sat, 23 Dec 2017 21:25:06 +0000 (16:25 -0500)]
Add ATTRIBUTE_PRINTF to printf_field_type_assignment
Get rid of:
/home/simark/src/binutils-gdb/gdb/target-descriptions.c:2026:25: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
vprintf_unfiltered (fmt, args);
Simon Marchi [Thu, 21 Dec 2017 23:19:17 +0000 (18:19 -0500)]
Do not emit "field_type" var if not needed on "maint print c-tdesc"
While fiddling a bit with -Wunused-variable, Sergio noticed that "maint
print c-tdesc" was always generating code for the "tdesc_type
*field_type" variable, even when it wasn't used. This is caught by GCC
when using -Wunused-variable, of course.
This patch changes the print_c_tdesc class to only output the field
declaration when we actually need it.
It shouldn't be necessary to do the same with the other variable
declarations (type_with_fields and element_type), because they are
always if they are declared.
The C files in features/ are regenerated, some declarations of
field_type are removed, as expected, while some others move to where
they are used for the first time.
opcodes/
* riscv-opc.c (match_c_add_with_hint, match_c_lui_with_hint): New.
(riscv_opcodes) <li>: Delete "d,0" line. Change Cj to Co.
<andi, and, add, addiw, addw, c.addi>: Change Cj to Co.
<add>: Add explanatory comment for 4-operand add instruction.
<c.nop>: Add support for immediate operand.
<c.mv, c.add>: Use match_c_add_with_hint instead of match_c_add.
<c.lui>: Use match_c_lui_with_hint instead of match_c_lui.
<c.li, c.slli>: Use match_opcode instead of match_rd_nonzero.
Ulrich Weigand [Wed, 20 Dec 2017 12:59:14 +0000 (13:59 +0100)]
[Cell/B.E.] Fix regression due to gdbarch_significant_addr_bit
On Cell/B.E. multi-architecture debugging we use a "merged" address space
that encodes both the main PowerPC address space and the local store address
spaces of all active SPUs. This will always occupy 64 bits.
However, gdbarch_addr_bit is set to 32 on SPU, and may be set to 32 as well
on PowerPC. Since the new gdbarch_significant_addr_bit defaults to the
value of gdbarch_addr_bit, this means addresses may be improperly truncated.
Work around this problem by explicitly setting gdbarch_significant_addr_bit
to 64 both for the SPU target and also for PowerPC target that support
Cell/B.E. execution.
Tamar Christina [Tue, 19 Dec 2017 12:05:20 +0000 (12:05 +0000)]
Correct disassembly of dot product instructions.
Dot products deviate from the normal disassembly rules for lane indexed
instruction. Their canonical representation is in the form of:
v0.2s, v0.8b, v0.4b[0] instead of v0.2s, v0.8b, v0.b[0] to try to denote
that these instructions select 4x 1 byte elements instead of a single 1 byte
element.
Previously we were disassembling them following the normal rules, this patch
corrects the disassembly.
Tamar Christina [Tue, 19 Dec 2017 12:04:13 +0000 (12:04 +0000)]
Add support for V_4B so we can properly reject it.
Previously parse_vector_type_for_operand was changed to allow the use of 4b
register size for indexed lane instructions. However this had the unintended
side effect of also allowing 4b for normal vector registers.
Because this support was only partial the rest of the tool silently treated
4b as 8b and continued. This patch adds full support for 4b so it can be
properly distinguished from 8b and the correct errors are generated.
With this patch you still can't encode any instruction which actually requires
v<num>.4b but such instructions don't exist so to prevent needing a workaround
in get_vreg_qualifier_from_value this was just omitted.
Nick Clifton [Tue, 19 Dec 2017 09:48:40 +0000 (09:48 +0000)]
Stop readelf from complaining about relocation sections with an sh_info field of 0 in dynamic executables.
PR 22587
* readelf.c (process_section_headers): Do not complain about an
sh_info field of 0 in relocation sections of ET_EXEC or ET_DYN
type executables.
Alan Modra [Mon, 18 Dec 2017 11:57:08 +0000 (22:27 +1030)]
PR22626, invalid dynindx used for dynamic relocs against section syms
_bfd_elf_link_renumber_dynsyms is called twice by the linker. The
first call in bfd_elf_size_dynamic_sections is just to answer the
question as to whether there are there any dynamic symbols. The
second call in bfd_elf_size_dynsym_hash_dynstr sets the st_shndx value
that dynamic symbols will have. strip_excluded_output_sections is
called between these two calls. So sections seen on the first
_bfd_elf_link_renumber_dynsyms pass might differ from those seen on
the second pass. Unfortunately, that can result in a stripped
section's dynamic symbol being assigned a dynindx on the first pass
but not corrected to the final value (of zero, ie. not dynamic) on the
second pass. PowerPC, x86, mips, and most other targets that emit
dynamic section symbols, just test that section symbol dynindx is
non-zero before using a given section symbol in dynamic relocations.
This patch prevents _bfd_elf_link_renumber_dynsyms from setting any
section symbol dynindx on the first pass.
PR 22626
* elflink.c (_bfd_elf_link_renumber_dynsyms): Don't set section
dynindx when section_sym_count is NULL.
(bfd_elf_size_dynamic_sections): Pass NULL section_sym_count to
preliminary _bfd_elf_link_renumber_dynsyms call.
Alan Modra [Fri, 15 Dec 2017 06:38:56 +0000 (17:08 +1030)]
Test binutils_assemble return value
This is a followup to "binutils nm testsuite tidy". Since the perror
in binutils_assemble has been removed, we need to take more care in
failure paths.
The patch also fixed a number of .exp files that have multiple tests,
where an assembly failure returns from the .exp file. In most cases
it is nicer to attempt all tests.
* testsuite/binutils-all/ar.exp (unique_symbol): Don't run AR
if assembly fails.
* testsuite/binutils-all/arc/objdump.exp (check_assembly): If
objfile is empty, fail test.
* testsuite/binutils-all/arm/objdump.exp: Don't return if assembly
fails for a test, continue on to other tests.
* testsuite/binutils-all/bfin/objdump.exp: Likewise.
* testsuite/binutils-all/hppa/objdump.exp: Likewise.
* testsuite/binutils-all/m68k/objdump.exp: Likewise.
* testsuite/binutils-all/vax/objdump.exp: Likewise.
* testsuite/binutils-all/size.exp: Likewise.
* testsuite/binutils-all/nm.exp: Likewise. Move PR12753 test.
* testsuite/binutils-all/objcopy.exp: Don't perror on assembly fail.
* testsuite/binutils-all/objdump.exp: Report assembly fails.
Nick Clifton [Mon, 18 Dec 2017 17:57:18 +0000 (17:57 +0000)]
Decode expression lists used with DW_AT_byte_size and other DWARF attributes.
PR 22532
* dwarf.c (read_and_display_attr_value): Add attributes that might
use the DW_FORM_exprloc form, and if so, display the decoded
location expression list.
Jan Beulich [Mon, 18 Dec 2017 08:37:15 +0000 (09:37 +0100)]
x86: fold certain AVX and AVX2 templates
Just like for instructions in GPRs, there's no need to have separate
templates for otherwise identical insns acting on XMM or YMM registers
(or memory of the same size).
Jan Beulich [Mon, 18 Dec 2017 08:34:00 +0000 (09:34 +0100)]
x86: replace Reg8, Reg16, Reg32, and Reg64
Use a combination of a single new Reg bit and Byte, Word, Dword, or
Qword instead.
Besides shrinking the number of operand type bits this has the benefit
of making register handling more similar to accumulator handling (a
generic flag is being accompanied by a "size qualifier"). It requires,
however, to split a few insn templates, as it is no longer correct to
have combinations like Reg32|Reg64|Byte. This slight growth in size will
hopefully be outweighed by this change paving the road for folding a
presumably much larger number of templates later on.
Joel Brobecker [Mon, 18 Dec 2017 03:39:33 +0000 (22:39 -0500)]
improved error message when getting an exception printing a variable
Consider the following Ada code defining a global variable whose
type is an array of static bounds (1 .. 2), but where its elements
are a variant record whose size is not statically known:
type Ints is array (Natural range <>) of Integer;
type Bounded_Ints (Max_Size : Natural) is record
Length : Natural := 0;
Objs : Ints (1 .. Max_Size);
end record;
type Ints_Doubled is array (1 .. 2) of Bounded_Ints (Idem (0));
Global : Ints_Doubled;
When compiling this program at -O2 using a GCC-6.4-based compiler
on x86_64-linux, trying to print the value of that global variable
yields:
(gdb) p global
$1 =
Let's look at the debugging info, which starts with the global
variable itself...
The error occurs while Ada evaluator is trying to "fix"
the element type inside the array, so as to determine its actual
size. For that, it searches for a parallel "XVZ" variable,
which, when found, contains the object's actual size.
Unfortunately in our case, the variable exists but has been
optimized out, as seen by the presence of a variable DIE in
the debugging info, but with no address attribute:
Discussing this with some members of AdaCore's compiler team,
it is expected that the optimizer can get rid of this variable,
and we don't want to pessimize the code just to improve debuggability,
since -O2 is about performance. So, the idea of this patch is
not to make it work, but provide a bit more information to help
users understand what kind of error is preventing GDB from being
able to print the variable's value.
The first hurdle we had to clear was the fact that ada_val_print
traps all exceptions (including QUIT ones!), and does so completly
silently. So, the fix was to add a trace of the exception being
generated. While doing so, we fix an old XXX/FIXME by only catching
errors, letting QUIT exceptions go through.
Once this is done, we now get an error message, which gives a first
clue as to what was happening:
(gdb) p fd.global
$1 = <error reading variable: value has been optimized out>
However, it would be more useful to know which value it was
that was optimized out. For that purpose, we enhanced
ada-lang.c::ada_to_fixed_type_1 so as to re-throw the error
with a message which indicates which variable we failed to read.
With those changes, the new output is now:
(gdb) p fd.global
$1 = <error reading variable: unable to read value of fd__Tints_doubledC___XVZ (value has been optimized out)>
gdb/ChangeLog:
* ada-lang.c (ada_to_fixed_type_1): Rethrow errors with
a more detailed exception message when getting an exception
while trying to read the value of an XVZ variable.
* ada-valprint.c (ada_val_print): Only catch RETURN_MASK_ERROR
exceptions. Print an error message when an exception is caught.
gdb/testsuite/ChangeLog:
* gdb.dwarf2/ada-valprint-error.c: New file.
* gdb.dwarf2/ada-valprint-error.exp: New file.
Joel Brobecker [Mon, 18 Dec 2017 03:09:27 +0000 (22:09 -0500)]
(Ada) crash assigning to record component which is an array
Consider the following code, which declares a variabled called "input"
of type "parameter", which is a record with one component called "u2",
where the type of that component is a simple 3-element array of
floating point values:
type Float_Array_3 is array (1 .. 3) of Float;
type parameters is record
u2 : Float_Array_3;
end record;
input : parameters;
Trying to assign a value to input.u2 causes GDB to crash:
Here, lhs_type is a TYPE_CODE_TYPEDEF. ada_is_direct_array_type
knows how to handle it, but TYPE_ARRAY_LOWER_BOUND_VALUE assumes
that the given type is a TYPE_CODE_ARRAY. As such, it ends up
accessing some fields in lhs_type which it shouldn't, and kaboom.
We fixed this issue by making sure that the TYPE_CODE_TYPEDEF
layer gets stripped.
Once this is done, we hit a different kind of error, also leading to
a SEGV, this time in assign_component. The code looks like this:
if (TYPE_CODE (value_type (lhs)) == TYPE_CODE_ARRAY)
[...]
else
[...]
Because once again lhs is a TYPE_CODE_TYPEDEF, the check fail,
and we end up assuming that lhs is a struct, executing the "else"
block, which is:
Xavier Roirand [Mon, 18 Dec 2017 02:59:07 +0000 (21:59 -0500)]
Ada: fix bad handling in ada_convert_actual
Using this small example:
procedure Foo is
type Integer_Access is access all Integer;
procedure P (A : Integer_Access) is
begin
null;
end P;
begin
P (null);
end Foo;
and doing this debug session:
(gdb) b p
Breakpoint 1 at 0x402d67: file foo.adb, line 7.
(gdb) print p(null)
Breakpoint 1, foo.p (a=0x641010) at foo.adb:10
... ^^^^^^^^^^
shows that something goes wrong between the initial null value and the
received parameter value in the 'f' function.
The value for the parameter 'a' we get is the address of the value we
would expect instead of the value itself. This can be checked by doing:
(gdb) p *a
$1 = 0
Before this fix, in ada_convert_value, this function was looking to the
actual value (the null value here) to determine if the formal (parameter
'a' in the procedure 'P' in this exemple) requires a pointer or not which
is a wrong assumption and leads to push the address of the value to the
inferior instead of the value itself.
This is fixed by this patch.
gdb/ChangeLog:
* ada-lang.c (ada_convert_actual): Change the way actual value
are passed to the inferior when the inferior expects a pointer type.
This commit implements the pahole-like '/o' option for 'ptype', which
prints the offsets and sizes of struct fields, reporting whenever
there is a hole found.
The output is heavily based on pahole(1), with a few modifications
here and there to adjust it to our reality. Here's an example:
A big part of this patch handles the formatting logic of 'ptype',
which is a bit messy. The code to handle bitfield offsets, however,
took some time to craft. My thanks to Pedro Alves for figuring things
out and pointing me to the right direction, as well as coming up with
a way to inspect the layout of structs with bitfields (see testcase
for comments).
After many discussions both on IRC and at the mailing list, I tried to
implement printing vtables and inherited classes. Unfortunately the
code grew too complex and there were still a few corner cases failing
so I had to drop the attempt. This should be implemented in a future
patch.
This patch is the start of a long-term work I'll do to flush the local
patches we carry for Fedora GDB. In this specific case, I'm aiming at
upstreaming the feature implemented by the 'pahole.py' script that is
shipped with Fedora GDB:
This has been regression-tested on the BuildBot. There's a new
testcase for it, along with an update to the documentation. I also
thought it was worth mentioning this feature in the NEWS file.
gdb/ChangeLog:
2017-12-15 Sergio Durigan Junior <sergiodj@redhat.com>
Pedro Alves <palves@redhat.com>
PR cli/16224
* NEWS (Changes since GDB 8.0): Mention new '/o' flag.
* c-typeprint.c (OFFSET_SPC_LEN): New define.
(c_type_print_varspec_prefix): New argument 'struct
print_offset_data *'.
(c_type_print_base_1): New function and prototype.
(c_print_type_1): New function, with code from 'c_print_type'.
(c_print_type): Use 'c_print_type_1'.
(c_type_print_varspec_prefix): New argument 'struct
print_offset_data *'. Use it. Call 'c_type_print_base_1'
instead of 'c_print_type_base'.
(print_spaces_filtered_with_print_options): New function.
(output_access_specifier): Take new argument FLAGS. Modify
function to call 'print_spaces_filtered_with_print_options'.
(c_print_type_vtable_offset_marker): New function.
(c_print_type_union_field_offset): New function.
(c_print_type_struct_field_offset): New function.
(c_print_type_no_offsets): New function.
(c_type_print_base_struct_union): New argument 'struct
print_offset_data *'. Print offsets and sizes for
struct/union/class fields.
* typeprint.c (const struct type_print_options
type_print_raw_options): Initialize 'print_offsets'.
(static struct type_print_options default_ptype_flags):
Likewise.
(struct print_offset_data print_offset_default_data): New
variable.
(whatis_exp): Handle '/o' option.
(_initialize_typeprint): Add '/o' flag to ptype's help.
* typeprint.h (struct print_offset_data): New struct.
(struct type_print_options) <print_offsets>: New field.
Reorganize code to handle TYPE_CODE_{STRUCT,UNION} on 'c_type_print_base'
While doing the 'ptype /o' work, I noticed that 'c_type_print_base'
was very long, with a big amount of code just to handle the case of
TYPE_CODE_{STRUCT,UNION}. This made working with the function a bit
difficult, specially because of the level of indentation.
This commit moves this part of the code to their own functions. Now
we have a 'c_type_print_base_struct_union' with most of the code, and
also 'need_access_label_p', which is a subset of the code that was
also a good candidate for having its own function.
* c-typeprint.c (need_access_label_p): New function.
(c_type_print_base_struct_union): New function.
(c_type_print_base): Move code to handle
TYPE_CODE_{STRUCT,UNION} to the functions mentioned above.
Nick Clifton [Fri, 15 Dec 2017 18:55:39 +0000 (18:55 +0000)]
Update documentation regarding the bfd returned by bfd_openr_next_archived_file
PR 22571
* archive.c (bfd_openr_next_archived_file): Extend the
documentation to note that it is necessary to call
bfd_check_format on the rrturned bfd before using it.
Yao Qi [Fri, 15 Dec 2017 10:45:27 +0000 (10:45 +0000)]
Skip 'maintenance check xml-descriptions' if XML is disabled
I see the following test failure when gdb is configured without XML
support,
maintenance check xml-descriptions binutils-gdb/gdb/testsuite/../features
warning: Can not parse XML target description; XML support was disabled at compile time^M
Tested 29 XML files, 29 failed
(gdb) FAIL: gdb.gdb/unittest.exp: maintenance check xml-descriptions ${srcdir}/../features
gdb/testsuite:
2017-12-15 Yao Qi <yao.qi@linaro.org>
* gdb.gdb/unittest.exp: Skip 'maintenance check xml-descriptions'
if XML is disabled.
Yao Qi [Fri, 15 Dec 2017 10:45:27 +0000 (10:45 +0000)]
Skip parse_memory_map_tests if XML is disabled
I find a fail in gdb unit test when gdb is configured without XML
support.
warning: Can not parse XML memory map; XML support was disabled at compile time^M
Self test failed: self-test failed at ../../binutils-gdb/gdb/unittests/memory-map-selftests.c:65
...
Ran 31 unit tests, 1 failed^M
(gdb) FAIL: gdb.gdb/unittest.exp: maintenance selftest
gdb:
2017-12-15 Yao Qi <yao.qi@linaro.org>
* unittests/memory-map-selftests.c: Wrap test with HAVE_LIBEXPAT.
Jan Beulich [Fri, 15 Dec 2017 08:14:52 +0000 (09:14 +0100)]
x86: correct operand type checks
Again these look to be typos: No template currently allows for any two
(or all three) of RegXMM, RegYMM, and RegZMM in a single operand. Quite
clearly ! are missing, after the addition of which the checks for the
first and (if present) second operands also fully match up.
Jan Beulich [Fri, 15 Dec 2017 08:13:54 +0000 (09:13 +0100)]
x86: drop stray CheckRegSize uses
They are relevant only when multiple operands permit registers:
operand_type_register_match() returns true if either operand is not a
register one. IOW