x86/ACPI: correct off-by-1 in SGI MMCFG check
PATCH 1/1 x86/ACPI: correct off-by-1 in SGI MMCFG check
As supported by the printk() (deliberately made visible in context by also correcting a mis-indented return statement), "above 4GiB" is meant here. Avoid comparison with a constant to "escape" Misra rule 7.2 complaints. (Note however that even up-to-date Linux, which is where we "inherited" this code from, still uses the very same off-by-1 check.)
xen/arch/x86/x86_64/acpi_mmcfg.c +2 -2
@@ -50,7 +50,7 @@ static int __init acpi_mcfg_check_entry( {int year; -if (cfg->address <0xFFFFFFFF ) +if (cfg->address == (uint32_t )cfg->address)return 0 ;if (!strncmp (mcfg->header.oem_id,"SGI" ,3 ))
@@ -59,7 +59,7 @@ static int __init acpi_mcfg_check_entry(if (mcfg->header.revision >=1 &&dmi_get_date (DMI_BIOS_DATE , &year,NULL ,NULL ) && year >=2010 ) -return 0 ; +return 0 ; printk(KERN_ERR "MCFG region for %04x:%02x-%02x at %#"PRIx64 " (above 4GB) ignored\n",
Quoted 8 lines · Jan Beulich
As supported by the
printk()(deliberately made visible in context by also correcting a mis-indented return statement), "above 4GiB" is meant here. Avoid comparison with a constant to "escape" Misra rule 7.2 complaints. (Note however that even up-to-date Linux, which is where we "inherited" this code from, still uses the very same off-by-1 check.)Fixes 94ea0622c5b8 ("x86-64/mmcfg: relax base address restriction")Signed-off-by Jan Beulich <[email protected]>
Albeit I wonder how relevant those checks are anymore, TBH I would be quite tempted to just drop all this.
Thanks, Roger.