Re: [PATCH 18/37] gdbstub: Move syscalls.c out of common_ss

[Richard Henderson]

On 3/13/25 10:46, Pierrick Bouvier wrote:

On 3/12/25 20:44, Richard Henderson wrote:

Copy to libuser_ss and libsystem_ss.
This file uses semihosting/semihost.h, which has separate
implementations with and without CONFIG_USER_ONLY.

Signed-off-by: Richard Henderson 
---
  gdbstub/meson.build | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gdbstub/meson.build b/gdbstub/meson.build
index 0e8099ae9c..b25db86767 100644
--- a/gdbstub/meson.build
+++ b/gdbstub/meson.build
@@ -7,15 +7,15 @@
  # We build two versions of gdbstub, one for each mode
  libuser_ss.add(files(
    'gdbstub.c',
+  'syscalls.c',
    'user.c'
  ))
  libsystem_ss.add(files(
    'gdbstub.c',
+  'syscalls.c',
    'system.c'
  ))
-common_ss.add(files('syscalls.c'))
-
  # The user-target is specialised by the guest
  specific_ss.add(when: 'CONFIG_USER_ONLY', if_true: files('user-target.c'))


So it fixes an existing issue if I'm correct?


I think it currently works accidentally, because semihosting_get_target() expands 
differently for user vs system.  But we provide semihosting/stubs-all.c which has another 
copy of the user version.



r~

Re: [PATCH 18/37] gdbstub: Move syscalls.c out of common_ss

[Pierrick Bouvier]

On 3/12/25 20:44, Richard Henderson wrote:

Copy to libuser_ss and libsystem_ss.
This file uses semihosting/semihost.h, which has separate
implementations with and without CONFIG_USER_ONLY.

Signed-off-by: Richard Henderson 
---
  gdbstub/meson.build | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gdbstub/meson.build b/gdbstub/meson.build
index 0e8099ae9c..b25db86767 100644
--- a/gdbstub/meson.build
+++ b/gdbstub/meson.build
@@ -7,15 +7,15 @@
  # We build two versions of gdbstub, one for each mode
  libuser_ss.add(files(
'gdbstub.c',
+  'syscalls.c',
'user.c'
  ))
  
  libsystem_ss.add(files(

'gdbstub.c',
+  'syscalls.c',
'system.c'
  ))
  
-common_ss.add(files('syscalls.c'))

-
  # The user-target is specialised by the guest
  specific_ss.add(when: 'CONFIG_USER_ONLY', if_true: files('user-target.c'))


So it fixes an existing issue if I'm correct?
Maybe there was no consequence, but definitely a hole in our build system.

Reviewed-by: Pierrick Bouvier