RE: [PATCH 3/3] gdb/arm: Track msp and psp

[Torbjorn SVENSSON <torbjorn.svensson@st.com>]

ST Restricted


> -----Original Message-----
> From: Luis Machado <luis.machado@arm.com>
> Sent: den 17 maj 2022 17:53
> To: Yvan ROUX - foss <yvan.roux@foss.st.com>; gdb-
> patches@sourceware.org
> Cc: Torbjorn SVENSSON <torbjorn.svensson@st.com>
> Subject: Re: [PATCH 3/3] gdb/arm: Track msp and psp
> 
> On 5/16/22 15:00, Yvan Roux via Gdb-patches wrote:
> > For Arm Cortex-M33 with security extensions, there are 4 different
> > stacks pointers (msp_s, msp_ns, psp_s, psp_ns).  To be compatible
> 
> stacks -> stack
> 
> > with earlier Cortex-M derivates, the msp and psp register are
> 
> register -> registers
> 
> > aliases for one of the 4 real stack pointer registers.
> >
> > These are the combinations that exist:
> > sp -> msp -> msp_s
> > sp -> msp -> msp_ns
> > sp -> psp -> psp_s
> > sp -> psp -> psp_ns
> >
> > This means that when the GDB client is to show the value of "msp",
> > the value should always be equal to either "msp_s" or "msp_ns".
> > Same goes for "psp".
> >
> 
> As a general comment, we should include, in the commit message, the bug
> ticket link related to this problem:
> 
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29121
> 
> > Signed-off-by: Torbj�rn SVENSSON <torbjorn.svensson@st.com>
> > Signed-off-by: Yvan Roux <yvan.roux@foss.st.com>
> > ---
> >   gdb/arm-tdep.c | 18 ++++++++++++++++++
> >   1 file changed, 18 insertions(+)
> >
> > diff --git a/gdb/arm-tdep.c b/gdb/arm-tdep.c
> > index 4aa277f5bc8..b9c35bcdae6 100644
> > --- a/gdb/arm-tdep.c
> > +++ b/gdb/arm-tdep.c
> > @@ -288,6 +288,8 @@ struct arm_prologue_cache
> >
> >     /* Active stack pointer.  */
> >     int active_sp_regnum;
> > +  int active_msp_regnum;
> > +  int active_psp_regnum;
> 
> Is it possible to have 3 active SP registers at the same time? Are msp and psp
> really needed here or are we just not doing a good
> job of figuring out what SP maps to? (msp_s, msp_ns, psp_s, psp_ns)

Only one is "active", but you can view them all at the same time.
The problem without tracking msp and psp (without suffixes) is that a simple "info registers" will not have anything to show for these 2 registers:

The following was executed in secure context of a Cortex-M33.
info r
r0             0x30000000          805306368
r1             0x74                116
r2             0x300000e8          805306600
r3             0xc00021d           201327133
r4             0x0                 0
r5             0x0                 0
r6             0x0                 0
r7             0x30017fe0          805404640
r8             0xffffffff          -1
r9             0xffffffff          -1
r10            0xffffffff          -1
r11            0xffffffff          -1
r12            0xffffffff          -1
sp             0x30017fe0          0x30017fe0
lr             0xc00ef57           201387863
pc             0xc00dc6e           0xc00dc6e <main+6>
xpsr           0x61000000          1627389952
fpscr          0x0                 0
primask        0x0                 0
basepri        0x0                 0
faultmask      0x0                 0
control        0x0                 0
msp            0x30017fe0          0x30017fe0
psp            0x0                 0x0
msp_ns         0x0                 0x0
psp_ns         0xfffffffc          0xfffffffc
msp_s          0x30017fe0          0x30017fe0
psp_s          0x0                 0x0
msplim_s       0x0                 0x0
psplim_s       0x0                 0x0
msplim_ns      0x0                 0x0
psplim_ns      0x0                 0x0
primask_s      0x0                 0
basepri_s      0x0                 0
faultmask_s    0x0                 0
control_s      0x0                 0
primask_ns     0x0                 0
basepri_ns     0x0                 0
faultmask_ns   0x0                 0
control_ns     0x0                 0


As the registers are listed, they need to be tracked, or do you see any other way to give the right value for the register?


Kind regards,
Torbjörn