S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
/* PLT trampolines. s390x version.
|
2018-01-01 00:32:25 +00:00
|
|
|
Copyright (C) 2016-2018 Free Software Foundation, Inc.
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
This file is part of the GNU C Library.
|
|
|
|
|
|
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
|
|
|
|
modify it under the terms of the GNU Lesser General Public
|
|
|
|
|
License as published by the Free Software Foundation; either
|
|
|
|
|
version 2.1 of the License, or (at your option) any later version.
|
|
|
|
|
|
|
|
|
|
The GNU C Library is distributed in the hope that it will be useful,
|
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
|
Lesser General Public License for more details.
|
|
|
|
|
|
|
|
|
|
You should have received a copy of the GNU Lesser General Public
|
|
|
|
|
License along with the GNU C Library; if not, see
|
|
|
|
|
<http://www.gnu.org/licenses/>. */
|
|
|
|
|
|
|
|
|
|
/* The PLT stubs will call _dl_runtime_resolve/_dl_runtime_profile
|
|
|
|
|
* with the following linkage:
|
|
|
|
|
* r2 - r6 : parameter registers
|
|
|
|
|
* f0, f2, f4, f6 : floating point parameter registers
|
|
|
|
|
* v24, v26, v28, v30, v25, v27, v29, v31 : vector parameter registers
|
|
|
|
|
* 48(r15), 56(r15) : PLT arguments PLT1, PLT2
|
|
|
|
|
* 160(r15) : additional stack parameters
|
2018-08-10 07:07:43 +00:00
|
|
|
* The slightly tightened normal clobber rules for function calls apply:
|
|
|
|
|
* r0 : call saved (for __fentry__)
|
|
|
|
|
* r1 - r5 : call clobbered
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
* r6 - r13 : call saved
|
|
|
|
|
* r14 : return address (call clobbered)
|
|
|
|
|
* r15 : stack pointer (call saved)
|
|
|
|
|
* f0 - f7 : call clobbered
|
|
|
|
|
* f8 - f15 : call saved
|
|
|
|
|
* v0 - v7 : bytes 0-7 overlap with f0-f7: call clobbered
|
|
|
|
|
bytes 8-15: call clobbered
|
|
|
|
|
* v8 - v15 : bytes 0-7 overlap with f8-f15: call saved
|
|
|
|
|
bytes 8-15: call clobbered
|
|
|
|
|
* v16 - v31 : call clobbered
|
|
|
|
|
*/
|
|
|
|
|
|
2018-08-10 07:07:42 +00:00
|
|
|
#define CFA_OFF 160
|
|
|
|
|
#define FRAME_OFF CFA_OFF + FRAME_SIZE
|
|
|
|
|
#define V24_OFF -288
|
|
|
|
|
#define V25_OFF -272
|
|
|
|
|
#define V26_OFF -256
|
|
|
|
|
#define V27_OFF -240
|
|
|
|
|
#define V28_OFF -224
|
|
|
|
|
#define V29_OFF -208
|
|
|
|
|
#define V30_OFF -192
|
|
|
|
|
#define V31_OFF -176
|
2018-08-10 07:07:43 +00:00
|
|
|
#define R0_OFF -120
|
2018-08-10 07:07:42 +00:00
|
|
|
#define PLT1_OFF -112
|
|
|
|
|
#define PLT2_OFF -104
|
|
|
|
|
#define R2_OFF -96
|
|
|
|
|
#define R3_OFF -88
|
|
|
|
|
#define R4_OFF -80
|
|
|
|
|
#define R5_OFF -72
|
|
|
|
|
#define R14_OFF -64
|
|
|
|
|
#define R15_OFF -56
|
|
|
|
|
#define F0_OFF -48
|
|
|
|
|
#define F2_OFF -40
|
|
|
|
|
#define F4_OFF -32
|
|
|
|
|
#define F6_OFF -24
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.globl _dl_runtime_resolve
|
|
|
|
|
.type _dl_runtime_resolve, @function
|
|
|
|
|
cfi_startproc
|
|
|
|
|
.align 16
|
|
|
|
|
_dl_runtime_resolve:
|
2018-08-10 07:07:43 +00:00
|
|
|
stg %r0,CFA_OFF+R0_OFF(%r15)
|
|
|
|
|
cfi_offset (r0, R0_OFF)
|
2018-08-10 07:07:42 +00:00
|
|
|
stmg %r2,%r5,CFA_OFF+R2_OFF(%r15) # save registers
|
|
|
|
|
cfi_offset (r2, R2_OFF)
|
|
|
|
|
cfi_offset (r3, R3_OFF)
|
|
|
|
|
cfi_offset (r4, R4_OFF)
|
|
|
|
|
cfi_offset (r5, R5_OFF)
|
|
|
|
|
stmg %r14,%r15,CFA_OFF+R14_OFF(%r15)
|
|
|
|
|
cfi_offset (r14, R14_OFF)
|
|
|
|
|
cfi_offset (r15, R15_OFF)
|
|
|
|
|
std %f0,CFA_OFF+F0_OFF(%r15)
|
|
|
|
|
cfi_offset (f0, F0_OFF)
|
|
|
|
|
std %f2,CFA_OFF+F2_OFF(%r15)
|
|
|
|
|
cfi_offset (f2, F2_OFF)
|
|
|
|
|
std %f4,CFA_OFF+F4_OFF(%r15)
|
|
|
|
|
cfi_offset (f4, F4_OFF)
|
|
|
|
|
std %f6,CFA_OFF+F6_OFF(%r15)
|
|
|
|
|
cfi_offset (f6, F6_OFF)
|
|
|
|
|
lmg %r2,%r3,CFA_OFF+PLT1_OFF(%r15) # load args saved by PLT
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
lgr %r0,%r15
|
|
|
|
|
#ifdef RESTORE_VRS
|
2018-08-10 07:07:42 +00:00
|
|
|
# define FRAME_SIZE (CFA_OFF + 128)
|
|
|
|
|
aghi %r15,-FRAME_SIZE # create stack frame
|
|
|
|
|
cfi_adjust_cfa_offset (FRAME_SIZE)
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine push
|
|
|
|
|
.machine "z13"
|
2018-08-10 07:07:42 +00:00
|
|
|
vstm %v24,%v31,FRAME_OFF+V24_OFF(%r15) # save call-clobbered vr args
|
|
|
|
|
cfi_offset (v24, V24_OFF)
|
|
|
|
|
cfi_offset (v25, V25_OFF)
|
|
|
|
|
cfi_offset (v26, V26_OFF)
|
|
|
|
|
cfi_offset (v27, V27_OFF)
|
|
|
|
|
cfi_offset (v28, V28_OFF)
|
|
|
|
|
cfi_offset (v29, V29_OFF)
|
|
|
|
|
cfi_offset (v30, V30_OFF)
|
|
|
|
|
cfi_offset (v31, V31_OFF)
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine pop
|
|
|
|
|
#else
|
2018-08-10 07:07:42 +00:00
|
|
|
# define FRAME_SIZE CFA_OFF
|
|
|
|
|
aghi %r15,-FRAME_SIZE # create stack frame
|
|
|
|
|
cfi_adjust_cfa_offset (FRAME_SIZE)
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
#endif
|
|
|
|
|
stg %r0,0(%r15) # write backchain
|
|
|
|
|
brasl %r14,_dl_fixup # call _dl_fixup
|
|
|
|
|
lgr %r1,%r2 # function addr returned in r2
|
|
|
|
|
#ifdef RESTORE_VRS
|
|
|
|
|
.machine push
|
|
|
|
|
.machine "z13"
|
2018-08-10 07:07:42 +00:00
|
|
|
vlm %v24,%v31,FRAME_OFF+V24_OFF(%r15) # restore vector registers
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine pop
|
|
|
|
|
#endif
|
2018-08-10 07:07:42 +00:00
|
|
|
lmg %r14,%r15,FRAME_OFF+R14_OFF(%r15) # restore frame and registers
|
|
|
|
|
#undef FRAME_SIZE
|
|
|
|
|
cfi_def_cfa_offset (CFA_OFF)
|
|
|
|
|
ld %f0,CFA_OFF+F0_OFF(%r15)
|
|
|
|
|
ld %f2,CFA_OFF+F2_OFF(%r15)
|
|
|
|
|
ld %f4,CFA_OFF+F4_OFF(%r15)
|
|
|
|
|
ld %f6,CFA_OFF+F6_OFF(%r15)
|
|
|
|
|
lmg %r2,%r5,CFA_OFF+R2_OFF(%r15)
|
2018-08-10 07:07:43 +00:00
|
|
|
lg %r0,CFA_OFF+R0_OFF(%r15)
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
br %r1
|
|
|
|
|
cfi_endproc
|
|
|
|
|
.size _dl_runtime_resolve, .-_dl_runtime_resolve
|
2018-08-10 07:07:42 +00:00
|
|
|
#undef V24_OFF
|
|
|
|
|
#undef V25_OFF
|
|
|
|
|
#undef V26_OFF
|
|
|
|
|
#undef V27_OFF
|
|
|
|
|
#undef V28_OFF
|
|
|
|
|
#undef V29_OFF
|
|
|
|
|
#undef V30_OFF
|
|
|
|
|
#undef V31_OFF
|
2018-08-10 07:07:43 +00:00
|
|
|
#undef R0_OFF
|
2018-08-10 07:07:42 +00:00
|
|
|
#undef PLT1_OFF
|
|
|
|
|
#undef PLT2_OFF
|
|
|
|
|
#undef R2_OFF
|
|
|
|
|
#undef R3_OFF
|
|
|
|
|
#undef R4_OFF
|
|
|
|
|
#undef R5_OFF
|
|
|
|
|
#undef R14_OFF
|
|
|
|
|
#undef R15_OFF
|
|
|
|
|
#undef F0_OFF
|
|
|
|
|
#undef F2_OFF
|
|
|
|
|
#undef F4_OFF
|
|
|
|
|
#undef F6_OFF
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
|
|
|
|
|
#ifndef PROF
|
2018-08-10 07:07:42 +00:00
|
|
|
# define SIZEOF_STRUCT_LA_S390_64_REGS 200
|
|
|
|
|
# define REGS_OFF -360
|
|
|
|
|
# define R2_OFF -360
|
|
|
|
|
# define R3_OFF -352
|
|
|
|
|
# define R4_OFF -344
|
|
|
|
|
# define R5_OFF -336
|
|
|
|
|
# define R6_OFF -328
|
|
|
|
|
# define F0_OFF -320
|
|
|
|
|
# define F2_OFF -312
|
|
|
|
|
# define F4_OFF -304
|
|
|
|
|
# define F6_OFF -296
|
|
|
|
|
# define V24_OFF -288
|
|
|
|
|
# define V25_OFF -272
|
|
|
|
|
# define V26_OFF -256
|
|
|
|
|
# define V27_OFF -240
|
|
|
|
|
# define V28_OFF -224
|
|
|
|
|
# define V29_OFF -208
|
|
|
|
|
# define V30_OFF -192
|
|
|
|
|
# define V31_OFF -176
|
2018-08-10 07:07:43 +00:00
|
|
|
# define R0_OFF -144
|
2018-08-10 07:07:42 +00:00
|
|
|
# define R12_OFF -136
|
|
|
|
|
# define R14_OFF -128
|
|
|
|
|
# define FRAMESIZE_OFF -120
|
|
|
|
|
# define PLT1_OFF -112
|
|
|
|
|
# define PLT2_OFF -104
|
|
|
|
|
# define PREGS_OFF -96
|
|
|
|
|
# define RETVAL_OFF -88
|
|
|
|
|
# define RET_R2_OFF -88
|
|
|
|
|
# define RET_F0_OFF -80
|
|
|
|
|
# define RET_V24_OFF -72
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.globl _dl_runtime_profile
|
|
|
|
|
.type _dl_runtime_profile, @function
|
|
|
|
|
cfi_startproc
|
|
|
|
|
.align 16
|
|
|
|
|
_dl_runtime_profile:
|
2018-08-10 07:07:43 +00:00
|
|
|
stg %r0,CFA_OFF+R0_OFF(%r15)
|
|
|
|
|
cfi_offset (r0, R0_OFF)
|
2018-08-10 07:07:42 +00:00
|
|
|
stg %r12,CFA_OFF+R12_OFF(%r15) # r12 is used as backup of r15
|
|
|
|
|
cfi_offset (r12, R12_OFF)
|
|
|
|
|
stg %r14,CFA_OFF+R14_OFF(%r15)
|
|
|
|
|
cfi_offset (r14, R14_OFF)
|
|
|
|
|
lgr %r12,%r15 # backup stack pointer
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
cfi_def_cfa_register (12)
|
2018-08-10 07:07:42 +00:00
|
|
|
# define FRAME_SIZE (CFA_OFF + SIZEOF_STRUCT_LA_S390_64_REGS)
|
|
|
|
|
aghi %r15,-FRAME_SIZE # create stack frame:
|
|
|
|
|
stg %r12,0(%r15) # save backchain
|
2016-03-31 15:37:16 +00:00
|
|
|
|
2018-08-10 07:07:42 +00:00
|
|
|
stmg %r2,%r6,FRAME_OFF+R2_OFF(%r15) # save call-clobbered arg regs
|
|
|
|
|
cfi_offset (r2, R2_OFF) # + r6 needed as arg for
|
|
|
|
|
cfi_offset (r3, R3_OFF) # _dl_profile_fixup
|
|
|
|
|
cfi_offset (r4, R4_OFF)
|
|
|
|
|
cfi_offset (r5, R5_OFF)
|
|
|
|
|
cfi_offset (r6, R6_OFF)
|
|
|
|
|
std %f0,FRAME_OFF+F0_OFF(%r15)
|
|
|
|
|
cfi_offset (f0, F0_OFF)
|
|
|
|
|
std %f2,FRAME_OFF+F2_OFF(%r15)
|
|
|
|
|
cfi_offset (f2, F2_OFF)
|
|
|
|
|
std %f4,FRAME_OFF+F4_OFF(%r15)
|
|
|
|
|
cfi_offset (f4, F4_OFF)
|
|
|
|
|
std %f6,FRAME_OFF+F6_OFF(%r15)
|
|
|
|
|
cfi_offset (f6, F6_OFF)
|
|
|
|
|
# ifdef RESTORE_VRS
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine push
|
|
|
|
|
.machine "z13"
|
2018-08-10 07:07:42 +00:00
|
|
|
vstm %v24,%v31,FRAME_OFF+V24_OFF(%r15) # store call-clobbered
|
|
|
|
|
cfi_offset (v24, V24_OFF) # vr arguments
|
|
|
|
|
cfi_offset (v25, V25_OFF)
|
|
|
|
|
cfi_offset (v26, V26_OFF)
|
|
|
|
|
cfi_offset (v27, V27_OFF)
|
|
|
|
|
cfi_offset (v28, V28_OFF)
|
|
|
|
|
cfi_offset (v29, V29_OFF)
|
|
|
|
|
cfi_offset (v30, V30_OFF)
|
|
|
|
|
cfi_offset (v31, V31_OFF)
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine pop
|
2018-08-10 07:07:42 +00:00
|
|
|
# endif
|
|
|
|
|
lmg %r2,%r3,CFA_OFF+PLT1_OFF(%r12) # load arguments saved by PLT
|
|
|
|
|
lgr %r4,%r14 # return address as 3rd parameter
|
|
|
|
|
la %r5,FRAME_OFF+REGS_OFF(%r15) # struct La_s390_64_regs *
|
|
|
|
|
la %r6,CFA_OFF+FRAMESIZE_OFF(%r12) # long int * framesize
|
|
|
|
|
brasl %r14,_dl_profile_fixup # call resolver
|
|
|
|
|
lgr %r1,%r2 # function addr returned in r2
|
|
|
|
|
ld %f0,FRAME_OFF+F0_OFF(%r15) # restore call-clobbered
|
|
|
|
|
ld %f2,FRAME_OFF+F2_OFF(%r15) # arg fprs
|
|
|
|
|
ld %f4,FRAME_OFF+F4_OFF(%r15)
|
|
|
|
|
ld %f6,FRAME_OFF+F6_OFF(%r15)
|
|
|
|
|
# ifdef RESTORE_VRS
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine push
|
2018-08-10 07:07:42 +00:00
|
|
|
.machine "z13" # restore call-clobbered
|
|
|
|
|
vlm %v24,%v31,FRAME_OFF+V24_OFF(%r15)# arg vrs
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
.machine pop
|
2018-08-10 07:07:42 +00:00
|
|
|
# endif
|
|
|
|
|
lg %r0,CFA_OFF+FRAMESIZE_OFF(%r12) # load framesize
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
ltgr %r0,%r0
|
|
|
|
|
jnm 1f
|
2018-08-10 07:07:42 +00:00
|
|
|
# framesize < 0 means no
|
|
|
|
|
lmg %r2,%r6,FRAME_OFF+R2_OFF(%r15) # pltexit call, so we can do a
|
|
|
|
|
# tail call without copying the
|
|
|
|
|
# arg overflow area
|
|
|
|
|
lgr %r15,%r12 # remove stack frame
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
cfi_def_cfa_register (15)
|
2018-08-10 07:07:42 +00:00
|
|
|
lg %r14,CFA_OFF+R14_OFF(%r15) # restore registers
|
|
|
|
|
lg %r12,CFA_OFF+R12_OFF(%r15)
|
2018-08-10 07:07:43 +00:00
|
|
|
lg %r0,CFA_OFF+R0_OFF(%r15)
|
2018-08-10 07:07:42 +00:00
|
|
|
br %r1 # tail call
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
|
|
|
|
|
cfi_def_cfa_register (12)
|
2018-08-10 07:07:42 +00:00
|
|
|
1: la %r4,FRAME_OFF+REGS_OFF(%r15) # struct La_s390_64_regs *
|
|
|
|
|
stg %r4,CFA_OFF+PREGS_OFF(%r12)
|
|
|
|
|
jz 4f # framesize == 0 ?
|
|
|
|
|
aghi %r0,7 # align framesize to 8
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
nill %r0,0xfff8
|
2018-08-10 07:07:42 +00:00
|
|
|
slgr %r15,%r0 # make room for framesize bytes
|
|
|
|
|
stg %r12,0(%r15) # save backchain
|
|
|
|
|
la %r2,FRAME_OFF+REGS_OFF(%r15)
|
|
|
|
|
la %r3,CFA_OFF(%r12)
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
srlg %r0,%r0,3
|
2018-08-10 07:07:42 +00:00
|
|
|
3: mvc 0(8,%r2),0(%r3) # copy additional parameters
|
|
|
|
|
la %r2,8(%r2) # depending on framesize
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
la %r3,8(%r3)
|
|
|
|
|
brctg %r0,3b
|
2018-08-10 07:07:42 +00:00
|
|
|
4: lmg %r2,%r6,0(%r4) # load register parameters
|
|
|
|
|
basr %r14,%r1 # call resolved function
|
|
|
|
|
stg %r2,CFA_OFF+RET_R2_OFF(%r12) # store return values r2, f0
|
|
|
|
|
std %f0,CFA_OFF+RET_F0_OFF(%r12) # to struct La_s390_64_retval
|
|
|
|
|
# ifdef RESTORE_VRS
|
2016-03-31 15:37:16 +00:00
|
|
|
.machine push
|
|
|
|
|
.machine "z13"
|
2018-08-10 07:07:42 +00:00
|
|
|
vst %v24,CFA_OFF+RET_V24_OFF(%r12) # store return value v24
|
2016-03-31 15:37:16 +00:00
|
|
|
.machine pop
|
2018-08-10 07:07:42 +00:00
|
|
|
# endif
|
|
|
|
|
lmg %r2,%r4,CFA_OFF+PLT1_OFF(%r12) # r2, r3: args saved by PLT
|
|
|
|
|
# r4: struct La_s390_64_regs *
|
|
|
|
|
la %r5,CFA_OFF+RETVAL_OFF(%r12) # struct La_s390_64_retval *
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
brasl %r14,_dl_call_pltexit
|
|
|
|
|
|
2018-08-10 07:07:42 +00:00
|
|
|
lgr %r15,%r12 # remove stack frame
|
|
|
|
|
# undef FRAME_SIZE
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
cfi_def_cfa_register (15)
|
2018-08-10 07:07:42 +00:00
|
|
|
lg %r14,CFA_OFF+R14_OFF(%r15) # restore registers
|
|
|
|
|
lg %r12,CFA_OFF+R12_OFF(%r15)
|
2018-08-10 07:07:43 +00:00
|
|
|
lg %r0,CFA_OFF+R0_OFF(%r15)
|
2018-08-10 07:07:42 +00:00
|
|
|
lg %r2,CFA_OFF+RET_R2_OFF(%r15) # restore return values
|
|
|
|
|
ld %f0,CFA_OFF+RET_F0_OFF(%r15)
|
|
|
|
|
# ifdef RESTORE_VRS
|
2016-03-31 15:37:16 +00:00
|
|
|
.machine push
|
|
|
|
|
.machine "z13"
|
2018-08-10 07:07:42 +00:00
|
|
|
vl %v24,CFA_OFF+RET_V24_OFF(%r15) # restore return value v24
|
2016-03-31 15:37:16 +00:00
|
|
|
.machine pop
|
2018-08-10 07:07:42 +00:00
|
|
|
# endif
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
br %r14 # Jump back to caller
|
|
|
|
|
|
|
|
|
|
cfi_endproc
|
|
|
|
|
.size _dl_runtime_profile, .-_dl_runtime_profile
|
2018-08-10 07:07:42 +00:00
|
|
|
# undef SIZEOF_STRUCT_LA_S390_64_REGS
|
|
|
|
|
# undef REGS_OFF
|
|
|
|
|
# undef R2_OFF
|
|
|
|
|
# undef R3_OFF
|
|
|
|
|
# undef R4_OFF
|
|
|
|
|
# undef R5_OFF
|
|
|
|
|
# undef R6_OFF
|
|
|
|
|
# undef F0_OFF
|
|
|
|
|
# undef F2_OFF
|
|
|
|
|
# undef F4_OFF
|
|
|
|
|
# undef F6_OFF
|
|
|
|
|
# undef V24_OFF
|
|
|
|
|
# undef V25_OFF
|
|
|
|
|
# undef V26_OFF
|
|
|
|
|
# undef V27_OFF
|
|
|
|
|
# undef V28_OFF
|
|
|
|
|
# undef V29_OFF
|
|
|
|
|
# undef V30_OFF
|
|
|
|
|
# undef V31_OFF
|
2018-08-10 07:07:43 +00:00
|
|
|
# undef R0_OFF
|
2018-08-10 07:07:42 +00:00
|
|
|
# undef R12_OFF
|
|
|
|
|
# undef R14_OFF
|
|
|
|
|
# undef FRAMESIZE_OFF
|
|
|
|
|
# undef PLT1_OFF
|
|
|
|
|
# undef PLT2_OFF
|
|
|
|
|
# undef PREGS_OFF
|
|
|
|
|
# undef RETVAL_OFF
|
|
|
|
|
# undef RET_R2_OFF
|
|
|
|
|
# undef RET_F0_OFF
|
|
|
|
|
# undef RET_V24_OFF
|
S390: Save and restore fprs/vrs while resolving symbols.
On s390, no fpr/vrs were saved while resolving a symbol
via _dl_runtime_resolve/_dl_runtime_profile.
According to the abi, the fpr-arguments are defined as call clobbered.
In leaf-functions, gcc 4.9 and newer can use fprs for saving/restoring gprs
instead of saving them to the stack.
If gcc do this in one of the resolver-functions, then the floating point
arguments of a library-function are invalid for the first library-function-call.
Thus, this patch saves/restores the fprs around the resolving code.
The same could occur for vector registers. Furthermore an ifunc-resolver
could also clobber the vector/floating point argument registers.
Thus this patch provides the further variants _dl_runtime_resolve_vx/
_dl_runtime_profile_vx, which are used if the kernel claims, that
we run on a machine with vector registers.
Furthermore, if _dl_runtime_profile calls _dl_call_pltexit,
the pointers to inregs-/outregs-structs were setup invalid.
Now they point to the correct location in the stack-frame.
Before branching back to the caller, the return values are now
restored instead of containing the return values of the
_dl_call_pltexit() call.
On s390-32, an endless loop occurs if _dl_call_pltexit() should be called.
Now, this code-path branches to this function instead of just after the
preceding basr-instruction.
ChangeLog:
* sysdeps/s390/s390-32/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-32/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
if _dl_call_pltexit is called.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
* sysdeps/s390/s390-64/dl-trampoline.S: Include dl-trampoline.h twice
to create a non-vector/vector version for _dl_runtime_resolve and
_dl_runtime_profile. Move implementation to ...
* sysdeps/s390/s390-64/dl-trampoline.h: ... here.
(_dl_runtime_resolve) Save and restore fpr/vrs.
(_dl_runtime_profile) Save and restore vrs and fix some issues
* sysdeps/s390/s390-64/dl-machine.h: (elf_machine_runtime_setup):
Choose the correct resolver function if running on a machine with vx.
2016-03-31 15:37:16 +00:00
|
|
|
#endif
|
