From: Kyle McMartin <kmcmarti@redhat.com>
To: gdb-patches@sourceware.org
Subject: [PATCH] aarch64: detect atomic sequences like other ll/sc architectures
Date: Mon, 24 Mar 2014 16:11:00 -0000 [thread overview]
Message-ID: <20140324161056.GB23291@redacted.bos.redhat.com> (raw)
Add similar single-stepping over atomic sequences support like other
load-locked/store-conditional architectures (alpha, powerpc, arm, etc.)
do. Verified the decode_masked_match, and decode_bcond works against the
atomic sequences used in the Linux kernel atomic.h, and also gcc
libatomic. Thanks to Richard Henderson for feedback on my initial
attempt at this patch!
2014-03-23 Kyle McMartin <kyle@redhat.com>
* aarch64-tdep.c (aarch64_deal_with_atomic_sequence): New function.
(aarch64_gdbarch_init): Handle single stepping of atomic sequences
with aarch64_deal_with_atomic_sequence.
--- a/gdb/aarch64-tdep.c
+++ b/gdb/aarch64-tdep.c
@@ -2509,6 +2509,82 @@ value_of_aarch64_user_reg (struct frame_info *frame, const void *baton)
}
\f
+static int
+aarch64_deal_with_atomic_sequence (struct frame_info *frame)
+{
+ struct gdbarch *gdbarch = get_frame_arch (frame);
+ struct address_space *aspace = get_frame_address_space (frame);
+ enum bfd_endian byte_order = gdbarch_byte_order (gdbarch);
+ const int insn_size = 4;
+ const int atomic_sequence_length = 16; /* Instruction sequence length. */
+ CORE_ADDR pc = get_frame_pc (frame);
+ CORE_ADDR breaks[2] = { -1, -1 };
+ CORE_ADDR loc = pc;
+ CORE_ADDR closing_insn = 0;
+ uint32_t insn = read_memory_unsigned_integer (loc, insn_size, byte_order);
+ int index;
+ int insn_count;
+ int bc_insn_count = 0; /* Conditional branch instruction count. */
+ int last_breakpoint = 0; /* Defaults to 0 (no breakpoints placed). */
+
+ /* look for a load-exclusive to begin the sequence... */
+ if (!decode_masked_match(insn, 0x3fc00000, 0x08400000))
+ return 0;
+
+ for (insn_count = 0; insn_count < atomic_sequence_length; ++insn_count)
+ {
+ int32_t offset;
+ unsigned cond;
+
+ loc += insn_size;
+ insn = read_memory_unsigned_integer (loc, insn_size, byte_order);
+
+ /* look for a conditional branch to set a breakpoint on the destination. */
+ if (decode_bcond(loc, insn, &cond, &offset))
+ {
+
+ if (bc_insn_count >= 1)
+ return 0;
+
+ breaks[1] = loc + offset;
+
+ bc_insn_count++;
+ last_breakpoint++;
+ }
+
+ /* and the matching store-exclusive to close it. */
+ if (decode_masked_match(insn, 0x3fc00000, 0x08000000))
+ {
+ closing_insn = loc;
+ break;
+ }
+ }
+
+ /* didn't find a stxr to end the sequence... */
+ if (!closing_insn)
+ return 0;
+
+ loc += insn_size;
+ insn = read_memory_unsigned_integer (loc, insn_size, byte_order);
+
+ /* insert breakpoint at the end of the atomic sequence */
+ breaks[0] = loc;
+
+ /* check for duplicated breakpoints. also check for a breakpoint placed on
+ * the conditional branch destination isn't within the sequence. */
+ if (last_breakpoint &&
+ (breaks[1] == breaks[0] ||
+ (breaks[1] >= pc && breaks[1] <= closing_insn)))
+ last_breakpoint = 0;
+
+ /* insert the breakpoint at the end of the sequence, also possibly at the
+ conditional branch destination */
+ for (index = 0; index <= last_breakpoint; index++)
+ insert_single_step_breakpoint (gdbarch, aspace, breaks[index]);
+
+ return 1;
+}
+
/* Initialize the current architecture based on INFO. If possible,
re-use an architecture from ARCHES, which is a list of
architectures already created during this debugging session.
@@ -2624,6 +2700,8 @@ aarch64_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
set_gdbarch_breakpoint_from_pc (gdbarch, aarch64_breakpoint_from_pc);
set_gdbarch_cannot_step_breakpoint (gdbarch, 1);
set_gdbarch_have_nonsteppable_watchpoint (gdbarch, 1);
+ /* Handles single stepping of atomic sequences. */
+ set_gdbarch_software_single_step (gdbarch, aarch64_deal_with_atomic_sequence);
/* Information about registers, etc. */
set_gdbarch_sp_regnum (gdbarch, AARCH64_SP_REGNUM);
next reply other threads:[~2014-03-24 16:11 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-03-24 16:11 Kyle McMartin [this message]
2014-03-24 16:25 ` Pedro Alves
2014-03-24 16:28 ` Kyle McMartin
2014-03-24 16:57 ` Joel Brobecker
2014-03-24 18:00 ` Richard Earnshaw
2014-03-27 1:52 ` Kyle McMartin
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=20140324161056.GB23291@redacted.bos.redhat.com \
--to=kmcmarti@redhat.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