[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <88c01eea-9d0f-4cac-914b-50993cbbb3ce@zytor.com>
Date: Tue, 5 Nov 2024 14:27:50 -0800
From: "H. Peter Anvin" <hpa@...or.com>
To: "Woodhouse, David" <dwmw@...zon.co.uk>,
"peterz@...radead.org" <peterz@...radead.org>,
"kexec@...ts.infradead.org" <kexec@...ts.infradead.org>,
"jpoimboe@...nel.org" <jpoimboe@...nel.org>
Cc: "horms@...nel.org" <horms@...nel.org>, "x86@...nel.org" <x86@...nel.org>,
"bp@...en8.de" <bp@...en8.de>, "mingo@...hat.com" <mingo@...hat.com>,
"tglx@...utronix.de" <tglx@...utronix.de>,
"kai.huang@...el.com" <kai.huang@...el.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"kirill.shutemov@...ux.intel.com" <kirill.shutemov@...ux.intel.com>,
"nik.borisov@...e.com" <nik.borisov@...e.com>,
"dave.hansen@...ux.intel.com" <dave.hansen@...ux.intel.com>
Subject: Re: [RFC PATCH 6/7] x86/kexec: Debugging support: Dump registers on
exception
On 11/5/24 13:58, H. Peter Anvin wrote:
>
> Then you can just do the obvious (have your assembly stub point %rdi to
> the base of the register dump; set the frame order to whatever you'd
> like, except rip/err/exc, or reverse the order if you prefer by changing
> the loop):
>
> static inline __noreturn void die(void)
> {
> while (1)
> asm volatile("hlt");
> }
>
> void dump_register_frame(const unsigned long frame[])
> {
> static const char regnames[][5] = {
> "rax:", "rcx:", "rdx:", "rbx:",
> "rsp:", "rbp:", "rsi:", "rdi:",
> "r8: ", "r9: ", "r10:", "r11:",
> "r12:", "r13:", "r14:", "r15:",
> "cr2:", "Exc:", "Err:", "rip:"
> };
>
> for (size_t i = 0; i < ARRAY_SIZE(regnames); i++) {
> __putstr(regnames[i]);
> __puthex(frame[i]);
> __putstr("\n");
> }
>
> /* Only return from int3 */
> if (frame[17] != 3)
> die();
> }
>
I don't know if it is necessary, but you can do something like this to
make absolutely sure you don't end up with non-relative symbol references:
static inline __constfunc void * sym_addr(const void *sym)
{
void *addr;
asm("lea %c1(%%rip),%0" : "=r" (addr) : "i" (sym));
return addr;
}
Powered by blists - more mailing lists