On 2025-12-11 00:48, Andrew Cooper wrote:
On 10/12/2025 8:31 pm, Nicola Vetrini wrote:
On 2025-12-10 21:09, Nicola Vetrini wrote:
On 2025-12-10 19:30, Andrew Cooper wrote:
diff --git a/xen/include/xen/elfstructs.h
b/xen/include/xen/elfstructs.h
index eb6b87a823a8..8770e7454672 100644
--- a/xen/include/xen/elfstructs.h
+++ b/xen/include/xen/elfstructs.h
@@ -360,7 +360,7 @@ typedef struct {
} Elf64_Rela;
#define ELF64_R_SYM(info) ((info) >> 32)
-#define ELF64_R_TYPE(info) ((info) & 0xFFFFFFFF)
+#define ELF64_R_TYPE(info) ((uint32_t)(info))
Actually I think this doesn't build:
arch/x86/livepatch.c: In function ‘arch_livepatch_perform_rela’:
././include/xen/config.h:55:24: error: format ‘%lu’ expects argument
of type ‘long unsigned int’, but argument 3 has type ‘unsigned int’
[-Werror=format=]
55 | #define XENLOG_ERR "<0>"
| ^~~~~
arch/x86/livepatch.c:332:20: note: in expansion of macro ‘XENLOG_ERR’
332 | printk(XENLOG_ERR LIVEPATCH "%s: Unhandled
relocation %lu\n",
| ^~~~~~~~~~
arch/x86/livepatch.c:332:69: note: format string is defined here
332 | printk(XENLOG_ERR LIVEPATCH "%s: Unhandled
relocation %lu\n",
|
~~^
|
|
|
long unsigned int
| %u
the error location is a bit unclear, but the cast is the culprit
Yeah, I spotted that just as I heading out, and ran
https://gitlab.com/xen-project/hardware/xen-staging/-/pipelines/2207521982
instead.
I've swapped back to using 0xFFFFFFFFU. info ends up being long, and
the result of the expression needs to stay that way.
However, looking at the report for that, I still missed one. I've
folded in this hunk too.
diff --git a/xen/arch/x86/pv/emulate.c b/xen/arch/x86/pv/emulate.c
index 8c44dea12330..e741e686c1af 100644
--- a/xen/arch/x86/pv/emulate.c
+++ b/xen/arch/x86/pv/emulate.c
@@ -37,7 +37,7 @@ int pv_emul_read_descriptor(unsigned int sel, const
struct vcpu *v,
if ( !(desc.b & _SEGMENT_L) )
{
*base = ((desc.a >> 16) + ((desc.b & 0xff) << 16) +
- (desc.b & 0xff000000));
+ (desc.b & 0xff000000U));
*limit = (desc.a & 0xffff) | (desc.b & 0x000f0000);
if ( desc.b & _SEGMENT_G )
*limit = ((*limit + 1) << 12) - 1;
Makes sense, feel free to retain my R-by with these two changes.
~Andrew
--
Nicola Vetrini, B.Sc.
Software Engineer
BUGSENG (https://bugseng.com)
LinkedIn: https://www.linkedin.com/in/nicola-vetrini-a42471253
