From: Luis Machado <luisgpm@linux.vnet.ibm.com>
To: gdb-patches@sourceware.org
Subject: [PATCH] PPC - Stepping off breakpoints in non-stop mode
Date: Fri, 02 May 2008 20:26:00 -0000 [thread overview]
Message-ID: <1209753019.7131.29.camel@gargoyle> (raw)
[-- Attachment #1: Type: text/plain, Size: 380 bytes --]
Hi folks,
This is the ppc-specific code to step off breakpoints in non-stop mode.
The main code is the fixup function, responsible for making sure we have
the correct PC after a displaced stepping has been concluded.
It applies on top of Pedro and Jim's more general displaced stepping
patch.
Tested without regressions on PPC 32/64.
Is this OK? Comments?
Best regards,
Luis
[-- Attachment #2: ppc_displaced.diff --]
[-- Type: text/x-patch, Size: 5783 bytes --]
2008-05-02 Luis Machado <luisgpm@br.ibm.com>
* ppc-tdep.h: Define PPC_MAX_INSN_LEN, BRANCH_MASK, B_INSN, BC_INSN,
LWARX_MASK, LWARX_INSTRUCTION, LDARX_INSTRUCTION, STWCX_MASK,
STWCX_INSTRUCTION, STDCX_INSTRUCTION, BXL_INSN, BP_MASK and BP_INSN.
* rs6000-tdep.c (ppc_displaced_step_fixup): New function.
(deal_with_atomic_sequence): Update BC masks.
(rs6000_gdbarch_init): Init displaced stepping infra-structure.
Remove LWARX_MASK, LWARX_INSTRUCTION, LDARX_INSTRUCTION, STWCX_MASK,
STWCX_INSTRUCTION, STDCX_INSTRUCTION, BC_MASK and BC_INSTRUCTION.
Index: HEAD/gdb/ppc-tdep.h
===================================================================
--- HEAD.orig/gdb/ppc-tdep.h 2008-05-01 14:00:40.000000000 -0700
+++ HEAD/gdb/ppc-tdep.h 2008-05-02 10:52:53.000000000 -0700
@@ -261,10 +261,28 @@
PPC_NUM_REGS
};
+/* The length of the longest ppc instruction. */
+#define PPC_MAX_INSN_LEN (4)
/* Instruction size. */
#define PPC_INSN_SIZE 4
+/* Instruction masks used during single-stepping of atomic sequences. */
+#define LWARX_MASK 0xfc0007fe
+#define LWARX_INSTRUCTION 0x7c000028
+#define LDARX_INSTRUCTION 0x7c0000A8
+#define STWCX_MASK 0xfc0007ff
+#define STWCX_INSTRUCTION 0x7c00012d
+#define STDCX_INSTRUCTION 0x7c0001ad
+
+/* Instruction masks for displaced stepping. */
+#define BRANCH_MASK 0xfc000000
+#define BP_MASK 0xFC0007FE
+#define B_INSN 0x48000000
+#define BC_INSN 0x40000000
+#define BXL_INSN 0x4c000000
+#define BP_INSN 0x7C000008
+
/* Estimate for the maximum number of instrctions in a function epilogue. */
#define PPC_MAX_EPILOGUE_INSTRUCTIONS 52
Index: HEAD/gdb/rs6000-tdep.c
===================================================================
--- HEAD.orig/gdb/rs6000-tdep.c 2008-04-30 22:05:39.000000000 -0700
+++ HEAD/gdb/rs6000-tdep.c 2008-05-02 11:22:04.000000000 -0700
@@ -985,16 +985,66 @@
return little_breakpoint;
}
-
-/* Instruction masks used during single-stepping of atomic sequences. */
-#define LWARX_MASK 0xfc0007fe
-#define LWARX_INSTRUCTION 0x7c000028
-#define LDARX_INSTRUCTION 0x7c0000A8
-#define STWCX_MASK 0xfc0007ff
-#define STWCX_INSTRUCTION 0x7c00012d
-#define STDCX_INSTRUCTION 0x7c0001ad
-#define BC_MASK 0xfc000000
-#define BC_INSTRUCTION 0x40000000
+/* Fix up the state of registers and memory after having single-stepped
+ a displaced instruction. */
+void
+ppc_displaced_step_fixup (struct gdbarch *gdbarch,
+ struct displaced_step_closure *closure,
+ CORE_ADDR from, CORE_ADDR to,
+ struct regcache *regs)
+{
+ /* Since we use simple_displaced_step_copy_insn, our closure is a
+ copy of the instruction. */
+ unsigned int *insn = (unsigned int *) closure;
+ unsigned int opcode = (*insn & BRANCH_MASK);
+ int offset = 4; /* Default offset for non PC-relative instructions. */
+
+ if (debug_displaced)
+ fprintf_unfiltered (gdb_stdlog,
+ "displaced: (ppc) fixup (0x%s, 0x%s)\n",
+ paddr_nz (from), paddr_nz (to));
+
+
+ /* Handle PC-relative branch instructions. */
+ if ((opcode == B_INSN) || (opcode == BC_INSN) || (opcode == BXL_INSN))
+ {
+ /* LK bit Indicates whether we should set the link register to point
+ to the next instruction or not. */
+ char link_register_bit = (char) (*insn & 0x1);
+ unsigned long current_pc;
+
+ /* Read the current PC value after the instruction has been executed
+ in a displaced location. Calculate the offset to be applied to the
+ original PC value before the displaced stepping. */
+ regcache_cooked_read_unsigned (regs, gdbarch_pc_regnum (gdbarch), ¤t_pc);
+ offset = current_pc - to;
+
+ if (debug_displaced)
+ fprintf_unfiltered (gdb_stdlog,
+ "displaced: (ppc) branch instruction: 0x%x\n"
+ "displaced: (ppc) adjusted PC from %s to %s\n",
+ *insn, paddr_nz (current_pc), paddr_nz (from + offset));
+
+ if (opcode != BXL_INSN)
+ {
+ /* AA bit indicating whether this is an absolute addressing or
+ PC-relative. */
+ char absolute_addr_bit = (char) (*insn & 0x2);
+
+ if (!absolute_addr_bit)
+ regcache_cooked_write_unsigned (regs, gdbarch_pc_regnum (gdbarch), from + offset);
+ }
+
+ if (link_register_bit)
+ regcache_cooked_write_unsigned (regs, 67, from + 4);
+ }
+ /* Check for breakpoints in the inferior. If we've found one, place the PC
+ right at the breakpoint instruction. */
+ else if ((*insn & BP_MASK) == BP_INSN)
+ regcache_cooked_write_unsigned (regs, gdbarch_pc_regnum (gdbarch), from);
+ else
+ regcache_cooked_write_unsigned (regs, gdbarch_pc_regnum (gdbarch), from + offset);
+}
/* Checks for an atomic sequence of instructions beginning with a LWARX/LDARX
instruction and ending with a STWCX/STDCX instruction. If such a sequence
@@ -1032,7 +1082,7 @@
/* Assume that there is at most one conditional branch in the atomic
sequence. If a conditional branch is found, put a breakpoint in
its destination address. */
- if ((insn & BC_MASK) == BC_INSTRUCTION)
+ if ((insn & BRANCH_MASK) == BC_INSN)
{
if (bc_insn_count >= 1)
return 0; /* More than one conditional branch found, fallback
@@ -3821,6 +3871,17 @@
/* Put the _Decimal128 pseudo-registers after the SPE registers. */
tdep->ppc_dl0_regnum += 32;
+ /* Setup displaced stepping. */
+ set_gdbarch_displaced_step_copy_insn (gdbarch,
+ simple_displaced_step_copy_insn);
+ set_gdbarch_displaced_step_fixup (gdbarch, ppc_displaced_step_fixup);
+ set_gdbarch_displaced_step_free_closure (gdbarch,
+ simple_displaced_step_free_closure);
+ set_gdbarch_displaced_step_location (gdbarch,
+ displaced_step_at_entry_point);
+
+ set_gdbarch_max_insn_length (gdbarch, PPC_MAX_INSN_LEN);
+
return gdbarch;
}
next reply other threads:[~2008-05-02 18:30 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-05-02 20:26 Luis Machado [this message]
2008-05-19 15:20 ` Pedro Alves
2008-05-19 15:21 ` Luis Machado
2008-05-29 15:41 ` Luis Machado
2008-06-05 20:03 ` Daniel Jacobowitz
2008-06-06 16:00 ` Luis Machado
2008-06-24 14:59 ` Luis Machado
2008-06-24 18:08 ` Pedro Alves
2008-06-24 18:19 ` Luis Machado
2008-06-25 12:54 ` Joel Brobecker
2008-06-25 13:35 ` Luis Machado
2008-06-25 13:51 ` Joel Brobecker
2008-06-25 15:48 ` Luis Machado
2008-06-25 18:49 ` Joel Brobecker
2008-06-30 17:04 ` Luis Machado
2008-05-19 15:20 ` Luis Machado
2008-07-08 1:39 Jonathan Larmour
2008-07-08 3:53 ` Luis Machado
2008-07-08 14:59 ` Jonathan Larmour
2008-07-08 15:13 ` Luis Machado
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1209753019.7131.29.camel@gargoyle \
--to=luisgpm@linux.vnet.ibm.com \
--cc=gdb-patches@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox