On Wednesday, February 8, 2023 1:44:39 AM CST Greenman, Gregory wrote:
> On Fri, 2023-01-20 at 01:56 +0800, Aiden Leong wrote:
>
> > Fix a bug introduced by:
> > commit 32ed101aa140 ("iwlwifi: convert all Qu with Jf devices to the new
> > config table"), so now we pick the FIRST matching config.
> >
> > Signed-off-by: Aiden Leong <aiden.leong@xxxxxxxxx>
> > ---
> > drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/wireless/intel/iwlwifi/pcie/drv.c
> > b/drivers/net/wireless/intel/iwlwifi/pcie/drv.c
index
> > 99768d6a6032..05764eef15a7 100644
> > --- a/drivers/net/wireless/intel/iwlwifi/pcie/drv.c
> > +++ b/drivers/net/wireless/intel/iwlwifi/pcie/drv.c
> > @@ -1456,7 +1456,7 @@ iwl_pci_find_dev_info(u16 device, u16
> > subsystem_device,
if (!num_devices)
> > return NULL;
> >
> > - for (i = num_devices - 1; i >= 0; i--) {
> > + for (i = 0; i < num_devices; i++) {
> > const struct iwl_dev_info *dev_info =
> > &iwl_dev_info_table[i];
> > if (dev_info->device != (u16)IWL_CFG_ANY &&
>
>
> It failed or internal testing, so it's more complicated. To traverse this
> table
from the beginning to the end requires some changes to the table
> itself and the "goto" wasn't omitted by a mistake, but for a reason...
> For the device that you have (device id 0x4DF0, sub-device id 0x0244,
> right?)
is it enough to have the first fix (disable
> tx_with_siso_diversity)?
Hi Gregory,
That's exactly why I put a warning in previous emails.
My opinion will be a little different than yours in this situation.
1. We SHOULD traverse this table from top to bottom to keep our source tree as
clean as possible.
2. One simple option is to reverse every config items in this table so the
logic keep the same.
3. Your team(I assume Luca Coelho is your colleague) may need to provide
further explaination about the `goto` line, since each change in kernel should
have a reason.
4. 0x4DF0, 0x0244 is correct. The question is: Will Intel release products
with same pid+subID but differenct STEP/RF_TYPE/RF_ID etc? If so, pid+subID
won't be enough.
To sum up, there will be three patches:
1. This patch still fixes the BUG introduced by the `goto` change.
2. Patch 2 should be [PATCH 1/2] in previous email.
3. Patch 3 reverses every items in this table. Your team can fine-tune the
order of each items. I won't submit this patch.
If you like my ideas, please merge patch1&2 along with another ident fix patch.
BTW, it has been a month since the first email. I'd appreciate if you reply
soon.
Cheers,
AidenAttachment:
signature.asc
Description: This is a digitally signed message part.
