Le dimanche 07 juin 2015 à 18:03 +0100, Grant Likely a écrit : > On Tue, 28 Apr 2015 09:29:56 +0200 > , Paul Kocialkowski <contact@xxxxxxxx> > wrote: > > This grabs the serial number shown in cpuinfo from the serial-number devicetree > > property in priority. When booting with ATAGs (and without device-tree), the > > provided number is still shown instead. > > > > Signed-off-by: Paul Kocialkowski <contact@xxxxxxxx> > > One comment below, but otherwise: > > Acked-by: Grant Likely <grant.likely@xxxxxxxxxx> I'm afraid this was merged in Russell's tree already: http://ftp.arm.linux.org.uk/cgit/linux-arm.git/commit/?h=for-next&id=3f599875e5202986b350618a617527ab441bf206 Still, it might be useful to make another patch on top with your comments! > > --- > > arch/arm/include/asm/system_info.h | 1 + > > arch/arm/kernel/setup.c | 23 +++++++++++++++++++++-- > > 2 files changed, 22 insertions(+), 2 deletions(-) > > > > diff --git a/arch/arm/include/asm/system_info.h b/arch/arm/include/asm/system_info.h > > index 720ea03..3860cbd40 100644 > > --- a/arch/arm/include/asm/system_info.h > > +++ b/arch/arm/include/asm/system_info.h > > @@ -17,6 +17,7 @@ > > > > /* information about the system we're running on */ > > extern unsigned int system_rev; > > +extern const char *system_serial; > > extern unsigned int system_serial_low; > > extern unsigned int system_serial_high; > > extern unsigned int mem_fclk_21285; > > diff --git a/arch/arm/kernel/setup.c b/arch/arm/kernel/setup.c > > index 1d60beb..b501754 100644 > > --- a/arch/arm/kernel/setup.c > > +++ b/arch/arm/kernel/setup.c > > @@ -93,6 +93,9 @@ unsigned int __atags_pointer __initdata; > > unsigned int system_rev; > > EXPORT_SYMBOL(system_rev); > > > > +const char *system_serial; > > +EXPORT_SYMBOL(system_serial); > > + > > Is there any need to export this symbol? It's only used by built-in > code. Not by modules. I agree, but I thought it would be consistent with the way system_serial_high/low are exported. I'm not sure this was ever needed either, but generally, I guess it makes sense to export parameters that are passed to the kernel (it was through ATAGs before, now it's through device-tree, but the logic remains the same). > > unsigned int system_serial_low; > > EXPORT_SYMBOL(system_serial_low); > > > > @@ -821,8 +824,25 @@ arch_initcall(customize_machine); > > > > static int __init init_machine_late(void) > > { > > + struct device_node *root; > > + int ret; > > + > > if (machine_desc->init_late) > > machine_desc->init_late(); > > + > > + root = of_find_node_by_path("/"); > > + if (root) { > > + ret = of_property_read_string(root, "serial-number", > > + &system_serial); > > of_property_read_string() will not modify the argument on failure. If > system_serial is initialized to NULL, then the failure path is > unnecessary. Makes sense. > > + if (ret) > > + system_serial = NULL; > > + } > > Calls to of_find_node* functions increment the refcount for the node. > Need an of_node_put() here, or just use the of_root pointer directly. It > is safe to call of_property_read_string() with a NULL node pointer too, > it will fail gracefully. > > So, the whole thing can safely boil down to: > > of_property_read_string(of_root, "serial-number", &system_serial); > if (!system_serial) > system_serial = kasprintf(GFP_KERNEL, "%08x%08x", > system_serial_high, system_serial_low); Looks good to me, feel free to commit this change, on top of: http://ftp.arm.linux.org.uk/cgit/linux-arm.git/commit/?h=for-next&id=3f599875e5202986b350618a617527ab441bf206
Attachment:
signature.asc
Description: This is a digitally signed message part