Hi Linus,
On Wed, 19 Dec 2012 20:34:01 -0800, Linus Torvalds wrote:
> On Wed, Dec 19, 2012 at 1:33 PM, Jean Delvare <khali@xxxxxxxxxxxx> wrote:
> >
> > 6 files changed, 674 insertions(+), 440 deletions(-)
>
> This is not what I get when I pull. What's up?
This is not what I get when I pull from another host either. The file
count is correct, but I get 689 insertions and 455 deletions.
I investigated it and found that this is caused by the differences being
expressed slightly differently. For example:
--- a/drivers/hwmon/hwmon.c
+++ b/drivers/hwmon/hwmon.c
@@ -84,19 +84,21 @@ static void __init hwmon_pci_quirks(void)
/* Open access to 0x295-0x296 on MSI MS-7031 */
sb = pci_get_device(PCI_VENDOR_ID_ATI, 0x436c, NULL);
- if (sb &&
- (sb->subsystem_vendor == 0x1462 && /* MSI */
- sb->subsystem_device == 0x0031)) { /* MS-7031 */
+ if (sb) {
+ if (sb->subsystem_vendor == 0x1462 && /* MSI */
+ sb->subsystem_device == 0x0031) { /* MS-7031 */
+ pci_read_config_byte(sb, 0x48, &enable);
+ pci_read_config_word(sb, 0x64, &base);
- pci_read_config_byte(sb, 0x48, &enable);
- pci_read_config_word(sb, 0x64, &base);
-
- if (base == 0 && !(enable & BIT(2))) {
- dev_info(&sb->dev,
- "Opening wide generic port at 0x295\n");
- pci_write_config_word(sb, 0x64, 0x295);
- pci_write_config_byte(sb, 0x48, enable | BIT(2));
+ if (base == 0 && !(enable & BIT(2))) {
+ dev_info(&sb->dev,
+ "Opening wide generic port at 0x295\n");
+ pci_write_config_word(sb, 0x64, 0x295);
+ pci_write_config_byte(sb, 0x48,
+ enable | BIT(2));
+ }
}
+ pci_dev_put(sb);
}
#endif
}
on my side, became:
--- a/drivers/hwmon/hwmon.c
+++ b/drivers/hwmon/hwmon.c
@@ -84,19 +84,21 @@ static void __init hwmon_pci_quirks(void)
/* Open access to 0x295-0x296 on MSI MS-7031 */
sb = pci_get_device(PCI_VENDOR_ID_ATI, 0x436c, NULL);
- if (sb &&
- (sb->subsystem_vendor == 0x1462 && /* MSI */
- sb->subsystem_device == 0x0031)) { /* MS-7031 */
-
- pci_read_config_byte(sb, 0x48, &enable);
- pci_read_config_word(sb, 0x64, &base);
-
- if (base == 0 && !(enable & BIT(2))) {
- dev_info(&sb->dev,
- "Opening wide generic port at 0x295\n");
- pci_write_config_word(sb, 0x64, 0x295);
- pci_write_config_byte(sb, 0x48, enable | BIT(2));
+ if (sb) {
+ if (sb->subsystem_vendor == 0x1462 && /* MSI */
+ sb->subsystem_device == 0x0031) { /* MS-7031 */
+ pci_read_config_byte(sb, 0x48, &enable);
+ pci_read_config_word(sb, 0x64, &base);
+
+ if (base == 0 && !(enable & BIT(2))) {
+ dev_info(&sb->dev,
+ "Opening wide generic port at 0x295\n");
+ pci_write_config_word(sb, 0x64, 0x295);
+ pci_write_config_byte(sb, 0x48,
+ enable | BIT(2));
+ }
}
+ pci_dev_put(sb);
}
#endif
}
on the far side. So the diffstat difference comes from an extra blank
line being deleted and reinserted.
You'll notice that 674 - 440 == 689 - 455 so I presume the different
numbers for it87.c have the same explanation and the end result would
be correct.
At first I thought that this was caused by me using diff option -d when
generating the patch originally. However dropping it did change the
output but still did not lead to what I get when pulling from git.
So I can only suspect that the differences were somehow reformatted by
the git instance on kernel.org. Maybe different versions of git have a
different diffing algorithm? I don't even know if git has its own diff
implementation or if it relies on the system diff command. For what
it's worth, I am running the following:
git version 1.7.7
GNU diff version 3.1
Also note that the patches went into git with quiltimport, if that
makes a difference.
If you have suggestions, I'm all ears, but I think you can just take
this pull request as is.
Thanks,
--
Jean Delvare
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
