* GDB will set bp at a function for RISC-V when step over this function call
@ 2021-09-17 7:57 Ge, Lei
2021-09-19 2:04 ` Simon Marchi
0 siblings, 1 reply; 4+ messages in thread
From: Ge, Lei @ 2021-09-17 7:57 UTC (permalink / raw)
To: gdb; +Cc: Xin, Wenyan
Hi all,
I have a question about GDB for RISC-V. When step over a function call, GDB will set a bp at this function only for RISC-V. Do you think it is a bug? Thanks.
I use GDB 10.2. Below is test code. Line 9 is a function call of test0. When program stops at main.c:9, then next. From RSP log, we can see that gdb will set bp at the addr of test0 for RISC-V. For other arch, it will not.
1
2 int test0()
3 {
4 return 0;
5 }
6
7 int main (int argc, char * argv[])
8 {
9 test0();
10 return 0;
11 }
(gdb) disassemble test0
Dump of assembler code for function test0:
Sending packet: $m2aaaaaa5c0,40#3d...Packet received: 83b7c7a791c7172500000335a5a38297eff0bff885472300f400a26002644101828069bf411122e4000881473e85226441018280011106ec22e80010aa872330
0x0000002aaaaaa5e4 <+0>: addi sp,sp,-16
0x0000002aaaaaa5e6 <+2>: sd s0,8(sp)
0x0000002aaaaaa5e8 <+4>: addi s0,sp,16
0x0000002aaaaaa5ea <+6>: li a5,0
0x0000002aaaaaa5ec <+8>: mv a0,a5
0x0000002aaaaaa5ee <+10>: ld s0,8(sp)
0x0000002aaaaaa5f0 <+12>: addi sp,sp,16
0x0000002aaaaaa5f2 <+14>: ret
(gdb) disassemble main
Dump of assembler code for function main:
0x0000002aaaaaa5f4 <+0>: addi sp,sp,-32
0x0000002aaaaaa5f6 <+2>: sd ra,24(sp)
0x0000002aaaaaa5f8 <+4>: sd s0,16(sp)
0x0000002aaaaaa5fa <+6>: addi s0,sp,32
0x0000002aaaaaa5fc <+8>: mv a5,a0
Sending packet: $m2aaaaaa600,40#0b...Packet received: b4fe2326f4feeff0fffd81473e85e260426405618280397122f84af0171400001304447f171900001309497f3309894006fc26f44eec52e856e413593940630f
0x0000002aaaaaa5fe <+10>: sd a1,-32(s0)
0x0000002aaaaaa602 <+14>: sw a5,-20(s0)
0x0000002aaaaaa606 <+18>: jal ra,0x2aaaaaa5e4 <test0>
=> 0x0000002aaaaaa60a <+22>: li a5,0
0x0000002aaaaaa60c <+24>: mv a0,a5
0x0000002aaaaaa60e <+26>: ld ra,24(sp)
0x0000002aaaaaa610 <+28>: ld s0,16(sp)
0x0000002aaaaaa612 <+30>: addi sp,sp,32
0x0000002aaaaaa614 <+32>: ret
(gdb) n
Sending packet: $m3ff7fed012,2#c3...Packet received: 8280
Sending packet: $X3ff7fed012,2:\002\220#7a...Packet received: OK
Sending packet: $m2aaaaaa606,2#df...Packet received: eff0
Sending packet: $m2aaaaaa608,2#e1...Packet received: fffd
Sending packet: $m2aaaaaa606,2#df...Packet received: eff0
Sending packet: $m2aaaaaa608,2#e1...Packet received: fffd
Sending packet: $m2aaaaaa5c0,40#3d...Packet received: 83b7c7a791c7172500000335a5a38297eff0bff885472300f400a26002644101828069bf411122e4000881473e85226441018280011106ec22e80010aa872330
Sending packet: $m2aaaaaa5e4,2#11...Packet received: 4111
Sending packet: $m2aaaaaa5e4,2#11...Packet received: 4111
Sending packet: $X2aaaaaa5e4,2:\002\220#c8...Packet received: OK
Sending packet: $vCont;c:p206.-1#76...Packet received: T05swbreak:;02:60fbffff3f000000;20:e4a5aaaa2a000000;thread:p206.206;core:0;
Sending packet: $X2aaaaaa5e4,2:A\021#88...Packet received: OK
Sending packet: $m2aaaaaa5c0,40#3d...Packet received: 83b7c7a791c7172500000335a5a38297eff0bff885472300f400a26002644101828069bf411122e4000881473e85226441018280011106ec22e80010aa872330
Sending packet: $m2aaaaaa5e4,2#11...Packet received: 4111
Sending packet: $m2aaaaaa5e4,4#13...Packet received: 411122e4
Sending packet: $m2aaaaaa5e0,4#0f...Packet received: 828069bf
Sending packet: $g#67...Packet received: 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[552 bytes omitted]
Sending packet: $m2aaaaaa60a,4#0c...Packet received: 81473e85
Sending packet: $m2aaaaaa606,4#e1...Packet received: eff0fffd
Sending packet: $m3ffffffb40,40#5a...Packet received: 0100000000000000b0d2fff73f0000008067baaa2a000000c067baaa2a000000f8fcffff3f000000000000000100000016a6aaaa2a0000000c3beff73f000000
Sending packet: $m2aaaaaa60a,2#0a...Packet received: 8147
Sending packet: $m2aaaaaa60a,2#0a...Packet received: 8147
Sending packet: $X2aaaaaa60a,2:\002\220#c1...Packet received: OK
Sending packet: $vCont;c:p206.-1#76...Packet received: T05swbreak:;02:60fbffff3f000000;20:0aa6aaaa2a000000;thread:p206.206;core:0;
Sending packet: $m2aaaaaa60a,4#0c...Packet received: 02903e85
Sending packet: $m2aaaaaa606,4#e1...Packet received: eff0fffd
Sending packet: $g#67...Packet received: 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[552 bytes omitted]
Sending packet: $m3ffffffb40,40#5a...Packet received: 0100000000000000b0d2fff73f0000008067baaa2a00000080fbffff3f000000f8fcffff3f000000000000000100000016a6aaaa2a0000000c3beff73f000000
Sending packet: $X2aaaaaa60a,2:\201G#f7...Packet received: OK
Sending packet: $qXfer:threads:read::0,1000#92...Packet received: l<threads>\n<thread id="p206.206" core="0" name="test"/>\n</threads>\n
Sending packet: $X3ff7fed012,2:\202\200#ea...Packet received: OK
Sending packet: $m2aaaaaa60a,4#0c...Packet received: 81473e85
Sending packet: $m2aaaaaa606,4#e1...Packet received: eff0fffd
10 return 0;
(gdb)
Thanks,
Lei
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: GDB will set bp at a function for RISC-V when step over this function call
2021-09-17 7:57 GDB will set bp at a function for RISC-V when step over this function call Ge, Lei
@ 2021-09-19 2:04 ` Simon Marchi
2021-09-20 17:06 ` John Baldwin
0 siblings, 1 reply; 4+ messages in thread
From: Simon Marchi @ 2021-09-19 2:04 UTC (permalink / raw)
To: Ge, Lei, gdb; +Cc: Xin, Wenyan
On 2021-09-17 3:57 a.m., Ge, Lei wrote:
> Hi all,
>
> I have a question about GDB for RISC-V. When step over a function call, GDB will set a bp at this function only for RISC-V. Do you think it is a bug? Thanks.
Hi,
Probably because RISC-V uses software single step to implement
single instruction stepping. It implements the
gdbarch_software_single_step hook:
https://gitlab.com/gnutools/binutils-gdb/-/blob/master/gdb/riscv-tdep.c#L3923-3940
Note that this callback is only installed in riscv_fbsd_init_abi and
riscv_linux_init_abi, so is only effective when debugging a Linux or
FreeBSD target. So I suppose that if you were debugging a bare-metal
application, GDB would use hardware single step instead, not sure if
that is by design.
Simon
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: GDB will set bp at a function for RISC-V when step over this function call
2021-09-19 2:04 ` Simon Marchi
@ 2021-09-20 17:06 ` John Baldwin
2021-09-22 21:48 ` Jim Wilson
0 siblings, 1 reply; 4+ messages in thread
From: John Baldwin @ 2021-09-20 17:06 UTC (permalink / raw)
To: Simon Marchi, Ge, Lei, gdb; +Cc: Xin, Wenyan
On 9/18/21 7:04 PM, Simon Marchi via Gdb wrote:
> On 2021-09-17 3:57 a.m., Ge, Lei wrote:
>> Hi all,
>>
>> I have a question about GDB for RISC-V. When step over a function call, GDB will set a bp at this function only for RISC-V. Do you think it is a bug? Thanks.
>
> Hi,
>
> Probably because RISC-V uses software single step to implement
> single instruction stepping. It implements the
> gdbarch_software_single_step hook:
>
> https://gitlab.com/gnutools/binutils-gdb/-/blob/master/gdb/riscv-tdep.c#L3923-3940
>
> Note that this callback is only installed in riscv_fbsd_init_abi and
> riscv_linux_init_abi, so is only effective when debugging a Linux or
> FreeBSD target. So I suppose that if you were debugging a bare-metal
> application, GDB would use hardware single step instead, not sure if
> that is by design.
RISC-V doesn't (to my knowledge) have support for hardware single step and
should be using software single step on bare metal as well.
--
John Baldwin
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: GDB will set bp at a function for RISC-V when step over this function call
2021-09-20 17:06 ` John Baldwin
@ 2021-09-22 21:48 ` Jim Wilson
0 siblings, 0 replies; 4+ messages in thread
From: Jim Wilson @ 2021-09-22 21:48 UTC (permalink / raw)
To: John Baldwin; +Cc: Simon Marchi, Ge, Lei, gdb, Xin, Wenyan
On Mon, Sep 20, 2021 at 10:06 AM John Baldwin <jhb@freebsd.org> wrote:
> RISC-V doesn't (to my knowledge) have support for hardware single step and
> should be using software single step on bare metal as well.
>
There is hardware single-step/breakpoint/watchpoint support. However, in
old versions of the debug spec (e.g. HiFive Unleashed), this support was
only visible via jtag, and hence unusable to linux/freebsd/etc. But
openocd does have hardware single-step/breakpoint/watchport support which
does work. People doing bare metal debugging are expected to use openocd
with gdb. In new versions of the debug spec (e.g. HiFive Unmatched), this
support is now visible in machine mode, but we need opensbi and kernel
support before gdb can use it. I don't know of anyone working on that,
though I might try it if I ever find the time.
Jim
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-09-22 21:49 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-09-17 7:57 GDB will set bp at a function for RISC-V when step over this function call Ge, Lei
2021-09-19 2:04 ` Simon Marchi
2021-09-20 17:06 ` John Baldwin
2021-09-22 21:48 ` Jim Wilson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).