2018-07-24 05:28:58

by Yue Haibing

Subject: [PATCH net-next] bnxt_en: Fix logic of forward the VF MAC address to PF in bnxt_vf_validate_set_mac

Based on the comments,req->l2addr must match the VF MAC address
if firmware spec >= 1.2.2, mac_ok can be true.

Signed-off-by: YueHaibing <[email protected]>
---
drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c | 7 ++-----
1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
index a649108..7925964 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
@@ -954,12 +954,9 @@ static int bnxt_vf_validate_set_mac(struct bnxt *bp, struct bnxt_vf_info *vf)
if (ether_addr_equal((const u8 *)req->l2_addr, vf->mac_addr))
mac_ok = true;
} else if (is_valid_ether_addr(vf->vf_mac_addr)) {
- if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr))
+ if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr) &&
+ bp->hwrm_spec_code >= 0x10202)
mac_ok = true;
- } else if (bp->hwrm_spec_code < 0x10202) {
- mac_ok = true;
- } else {
- mac_ok = true;
}
if (mac_ok)
return bnxt_hwrm_exec_fwd_resp(bp, vf, msg_size);
--
2.7.0

2018-07-24 07:32:34

by Michael Chan

Subject: Re: [PATCH net-next] bnxt_en: Fix logic of forward the VF MAC address to PF in bnxt_vf_validate_set_mac

On Mon, Jul 23, 2018 at 10:24 PM, YueHaibing <[email protected]> wrote:
> Based on the comments,req->l2addr must match the VF MAC address
> if firmware spec >= 1.2.2, mac_ok can be true.
>
> Signed-off-by: YueHaibing <[email protected]>
> ---
> drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c | 7 ++-----
> 1 file changed, 2 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
> index a649108..7925964 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
> @@ -954,12 +954,9 @@ static int bnxt_vf_validate_set_mac(struct bnxt *bp, struct bnxt_vf_info *vf)
> if (ether_addr_equal((const u8 *)req->l2_addr, vf->mac_addr))
> mac_ok = true;
> } else if (is_valid_ether_addr(vf->vf_mac_addr)) {
> - if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr))
> + if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr) &&
> + bp->hwrm_spec_code >= 0x10202)
> mac_ok = true;

I'm not sure if this is correct. If firmware spec < 0x10202, the VF
MAC address is not forwarded to the PF and so it doesn't have to match
and mac_ok should still be true. I think we are missing that
condition with this patch.

I need to let my colleague Vasundhara comment on this. She is more
familiar with this logic.

> - } else if (bp->hwrm_spec_code < 0x10202) {
> - mac_ok = true;
> - } else {
> - mac_ok = true;
> }
> if (mac_ok)
> return bnxt_hwrm_exec_fwd_resp(bp, vf, msg_size);
> --
> 2.7.0
>
>

2018-07-24 16:02:38

by Vasundhara Volam

Subject: Re: [PATCH net-next] bnxt_en: Fix logic of forward the VF MAC address to PF in bnxt_vf_validate_set_mac

On Tue, Jul 24, 2018 at 1:01 PM, Michael Chan <[email protected]> wrote:
>
> On Mon, Jul 23, 2018 at 10:24 PM, YueHaibing <[email protected]> wrote:
> > Based on the comments,req->l2addr must match the VF MAC address
> > if firmware spec >= 1.2.2, mac_ok can be true.
> >
> > Signed-off-by: YueHaibing <[email protected]>
> > ---
> > drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c | 7 ++-----
> > 1 file changed, 2 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
> > index a649108..7925964 100644
> > --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
> > +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
> > @@ -954,12 +954,9 @@ static int bnxt_vf_validate_set_mac(struct bnxt *bp, struct bnxt_vf_info *vf)
> > if (ether_addr_equal((const u8 *)req->l2_addr, vf->mac_addr))
> > mac_ok = true;
> > } else if (is_valid_ether_addr(vf->vf_mac_addr)) {
> > - if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr))
> > + if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr) &&
> > + bp->hwrm_spec_code >= 0x10202)
> > mac_ok = true;
>
> I'm not sure if this is correct. If firmware spec < 0x10202, the VF
> MAC address is not forwarded to the PF and so it doesn't have to match
> and mac_ok should still be true. I think we are missing that
> condition with this patch.
>
> I need to let my colleague Vasundhara comment on this. She is more
> familiar with this logic.
Yes Michael, you are right. Also, the plain else condition is to cover
a special case to allow VF to modify
it's own MAC when PF has not assigned a valid MAC address and HWRM
spec code > 0x10202.
>
> > - } else if (bp->hwrm_spec_code < 0x10202) {
> > - mac_ok = true;
> > - } else {
> > - mac_ok = true;
> > }
> > if (mac_ok)
> > return bnxt_hwrm_exec_fwd_resp(bp, vf, msg_size);
> > --
> > 2.7.0
> >
> >

2018-07-24 21:49:43

by Michael Chan

Subject: Re: [PATCH net-next] bnxt_en: Fix logic of forward the VF MAC address to PF in bnxt_vf_validate_set_mac

On Tue, Jul 24, 2018 at 9:01 AM, Vasundhara Volam
<[email protected]> wrote:
> On Tue, Jul 24, 2018 at 1:01 PM, Michael Chan <[email protected]> wrote:
>>
>> On Mon, Jul 23, 2018 at 10:24 PM, YueHaibing <[email protected]> wrote:
>> > Based on the comments,req->l2addr must match the VF MAC address
>> > if firmware spec >= 1.2.2, mac_ok can be true.
>> >
>> > Signed-off-by: YueHaibing <[email protected]>
>> > ---
>> > drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c | 7 ++-----
>> > 1 file changed, 2 insertions(+), 5 deletions(-)
>> >
>> > diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
>> > index a649108..7925964 100644
>> > --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
>> > +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
>> > @@ -954,12 +954,9 @@ static int bnxt_vf_validate_set_mac(struct bnxt *bp, struct bnxt_vf_info *vf)
>> > if (ether_addr_equal((const u8 *)req->l2_addr, vf->mac_addr))
>> > mac_ok = true;
>> > } else if (is_valid_ether_addr(vf->vf_mac_addr)) {
>> > - if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr))
>> > + if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr) &&
>> > + bp->hwrm_spec_code >= 0x10202)
>> > mac_ok = true;
>>
>> I'm not sure if this is correct. If firmware spec < 0x10202, the VF
>> MAC address is not forwarded to the PF and so it doesn't have to match
>> and mac_ok should still be true. I think we are missing that
>> condition with this patch.
>>
>> I need to let my colleague Vasundhara comment on this. She is more
>> familiar with this logic.
> Yes Michael, you are right. Also, the plain else condition is to cover
> a special case to allow VF to modify
> it's own MAC when PF has not assigned a valid MAC address and HWRM
> spec code > 0x10202.

We should combine the "else if" and "else" below into a plain else and
add some comments to explain the conditions.

>>
>> > - } else if (bp->hwrm_spec_code < 0x10202) {
>> > - mac_ok = true;
>> > - } else {
>> > - mac_ok = true;
>> > }
>> > if (mac_ok)
>> > return bnxt_hwrm_exec_fwd_resp(bp, vf, msg_size);
>> > --
>> > 2.7.0
>> >
>> >

2018-07-25 10:20:24

by Yue Haibing

Subject: Re: [PATCH net-next] bnxt_en: Fix logic of forward the VF MAC address to PF in bnxt_vf_validate_set_mac

On 2018/7/25 5:48, Michael Chan wrote:
> On Tue, Jul 24, 2018 at 9:01 AM, Vasundhara Volam
> <[email protected]> wrote:
>> On Tue, Jul 24, 2018 at 1:01 PM, Michael Chan <[email protected]> wrote:
>>>
>>> On Mon, Jul 23, 2018 at 10:24 PM, YueHaibing <[email protected]> wrote:
>>>> Based on the comments,req->l2addr must match the VF MAC address
>>>> if firmware spec >= 1.2.2, mac_ok can be true.
>>>>
>>>> Signed-off-by: YueHaibing <[email protected]>
>>>> ---
>>>> drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c | 7 ++-----
>>>> 1 file changed, 2 insertions(+), 5 deletions(-)
>>>>
>>>> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
>>>> index a649108..7925964 100644
>>>> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
>>>> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c
>>>> @@ -954,12 +954,9 @@ static int bnxt_vf_validate_set_mac(struct bnxt *bp, struct bnxt_vf_info *vf)
>>>> if (ether_addr_equal((const u8 *)req->l2_addr, vf->mac_addr))
>>>> mac_ok = true;
>>>> } else if (is_valid_ether_addr(vf->vf_mac_addr)) {
>>>> - if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr))
>>>> + if (ether_addr_equal((const u8 *)req->l2_addr, vf->vf_mac_addr) &&
>>>> + bp->hwrm_spec_code >= 0x10202)
>>>> mac_ok = true;
>>>
>>> I'm not sure if this is correct. If firmware spec < 0x10202, the VF
>>> MAC address is not forwarded to the PF and so it doesn't have to match
>>> and mac_ok should still be true. I think we are missing that
>>> condition with this patch.
>>>
>>> I need to let my colleague Vasundhara comment on this. She is more
>>> familiar with this logic.
>> Yes Michael, you are right. Also, the plain else condition is to cover
>> a special case to allow VF to modify
>> it's own MAC when PF has not assigned a valid MAC address and HWRM
>> spec code > 0x10202.
>
> We should combine the "else if" and "else" below into a plain else and
> add some comments to explain the conditions.

Thank you for clarification.

I will send a new patch for this.

>
>>>
>>>> - } else if (bp->hwrm_spec_code < 0x10202) {
>>>> - mac_ok = true;
>>>> - } else {
>>>> - mac_ok = true;
>>>> }
>>>> if (mac_ok)
>>>> return bnxt_hwrm_exec_fwd_resp(bp, vf, msg_size);
>>>> --
>>>> 2.7.0
>>>>
>>>>
>
> .
>