Re: [PATCH 2/4] net: e1000: move all rx data structures to hw instance

2023-04-06 Thread Christian Gmeiner
>
> On Wed, Apr 05, 2023 at 09:50:08AM +0200, Christian Gmeiner wrote:
> > Hi all
> >
> > >
> > > Preparation for per driver instance allocated data structures.
> > >
> > > Signed-off-by: Christian Gmeiner 
> >
> > gentle ping - what can I do better to get some feedback, RBs, etc
> > faster? This experience is
> > quite frustrating and it happens for me from time to time for U-Boot 
> > patches :(
> >
> > Patch 1 of this series got a RB few days ago and the others are still open.
>
>
> Unfortunately there's a number of areas where we just don't have a lot
> of maintainers. Now that v2023.04 is out I expect to be picking up
> assorted patches that look right, soon.  Sorry for the delay.
>

Thanks for your reasoning. Out of this I started to review some k3 patches :)

-- 
greets
--
Christian Gmeiner, MSc

https://christian-gmeiner.info/privacypolicy


Re: [PATCH 2/4] net: e1000: move all rx data structures to hw instance

2023-04-05 Thread Tom Rini
On Wed, Apr 05, 2023 at 09:50:08AM +0200, Christian Gmeiner wrote:
> Hi all
> 
> >
> > Preparation for per driver instance allocated data structures.
> >
> > Signed-off-by: Christian Gmeiner 
> 
> gentle ping - what can I do better to get some feedback, RBs, etc
> faster? This experience is
> quite frustrating and it happens for me from time to time for U-Boot patches 
> :(
> 
> Patch 1 of this series got a RB few days ago and the others are still open.


Unfortunately there's a number of areas where we just don't have a lot
of maintainers. Now that v2023.04 is out I expect to be picking up
assorted patches that look right, soon.  Sorry for the delay.

-- 
Tom


signature.asc
Description: PGP signature


Re: [PATCH 2/4] net: e1000: move all rx data structures to hw instance

2023-04-05 Thread Christian Gmeiner
Hi all

>
> Preparation for per driver instance allocated data structures.
>
> Signed-off-by: Christian Gmeiner 

gentle ping - what can I do better to get some feedback, RBs, etc
faster? This experience is
quite frustrating and it happens for me from time to time for U-Boot patches :(

Patch 1 of this series got a RB few days ago and the others are still open.

> ---
>  drivers/net/e1000.c | 26 +++---
>  drivers/net/e1000.h |  5 +
>  2 files changed, 20 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/net/e1000.c b/drivers/net/e1000.c
> index 8dccf29c7e..08c84ce3d1 100644
> --- a/drivers/net/e1000.c
> +++ b/drivers/net/e1000.c
> @@ -61,10 +61,9 @@ tested on both gig copper and gig fiber boards
>   */
>  DEFINE_ALIGN_BUFFER(struct e1000_tx_desc, tx_base, 16, E1000_BUFFER_ALIGN);
>  DEFINE_ALIGN_BUFFER(struct e1000_rx_desc, rx_base, NUM_RX_DESC, 
> E1000_BUFFER_ALIGN);
> -DEFINE_ALIGN_BUFFER(unsigned char, packet, 4096, E1000_BUFFER_ALIGN);
> +DEFINE_ALIGN_BUFFER(unsigned char, _packet, 4096, E1000_BUFFER_ALIGN);
>
>  static int tx_tail;
> -static int rx_tail, rx_last;
>  static int num_cards;  /* Number of E1000 devices seen so far */
>
>  static struct pci_device_id e1000_supported[] = {
> @@ -5090,12 +5089,13 @@ e1000_sw_init(struct e1000_hw *hw)
>  void
>  fill_rx(struct e1000_hw *hw)
>  {
> +   unsigned char *packet = hw->rx_packet;
> struct e1000_rx_desc *rd;
> unsigned long flush_start, flush_end;
>
> -   rx_last = rx_tail;
> -   rd = rx_base + rx_tail;
> -   rx_tail = (rx_tail + 1) % NUM_RX_DESC;
> +   hw->rx_last = hw->rx_tail;
> +   rd = hw->rx_base + hw->rx_tail;
> +   hw->rx_tail = (hw->rx_tail + 1) % NUM_RX_DESC;
> memset(rd, 0, 16);
> rd->buffer_addr = cpu_to_le64(virt_to_phys(packet));
>
> @@ -5111,7 +5111,7 @@ fill_rx(struct e1000_hw *hw)
> flush_end = flush_start + roundup(sizeof(*rd), ARCH_DMA_MINALIGN);
> flush_dcache_range(flush_start, flush_end);
>
> -   E1000_WRITE_REG(hw, RDT, rx_tail);
> +   E1000_WRITE_REG(hw, RDT, hw->rx_tail);
>  }
>
>  /**
> @@ -5248,7 +5248,7 @@ static void
>  e1000_configure_rx(struct e1000_hw *hw)
>  {
> unsigned long rctl, ctrl_ext;
> -   rx_tail = 0;
> +   hw->rx_tail = 0;
>
> /* make sure receives are disabled while setting up the descriptors */
> rctl = E1000_READ_REG(hw, RCTL);
> @@ -5269,8 +5269,8 @@ e1000_configure_rx(struct e1000_hw *hw)
> E1000_WRITE_FLUSH(hw);
> }
> /* Setup the Base and Length of the Rx Descriptor Ring */
> -   E1000_WRITE_REG(hw, RDBAL, lower_32_bits(virt_to_phys(rx_base)));
> -   E1000_WRITE_REG(hw, RDBAH, upper_32_bits(virt_to_phys(rx_base)));
> +   E1000_WRITE_REG(hw, RDBAL, lower_32_bits(virt_to_phys(hw->rx_base)));
> +   E1000_WRITE_REG(hw, RDBAH, upper_32_bits(virt_to_phys(hw->rx_base)));
>
> E1000_WRITE_REG(hw, RDLEN, NUM_RX_DESC * sizeof(struct 
> e1000_rx_desc));
>
> @@ -5298,12 +5298,13 @@ POLL - Wait for a frame
>  static int
>  _e1000_poll(struct e1000_hw *hw)
>  {
> +   unsigned char *packet = hw->rx_packet;
> struct e1000_rx_desc *rd;
> unsigned long inval_start, inval_end;
> uint32_t len;
>
> /* return true if there's an ethernet packet ready to read */
> -   rd = rx_base + rx_last;
> +   rd = hw->rx_base + hw->rx_last;
>
> /* Re-load the descriptor from RAM. */
> inval_start = ((unsigned long)rd) & ~(ARCH_DMA_MINALIGN - 1);
> @@ -5469,6 +5470,9 @@ static int e1000_init_one(struct e1000_hw *hw, int 
> cardnum,
> return -EPERM;
> }
>
> +   hw->rx_base = rx_base;
> +   hw->rx_packet = _packet;
> +
> /* Are these variables needed? */
> hw->fc = e1000_fc_default;
> hw->original_fc = e1000_fc_default;
> @@ -5663,7 +5667,7 @@ static int e1000_eth_recv(struct udevice *dev, int 
> flags, uchar **packetp)
>
> len = _e1000_poll(hw);
> if (len)
> -   *packetp = packet;
> +   *packetp = hw->rx_packet;
>
> return len ? len : -EAGAIN;
>  }
> diff --git a/drivers/net/e1000.h b/drivers/net/e1000.h
> index 69882ba66f..f83e3a0b33 100644
> --- a/drivers/net/e1000.h
> +++ b/drivers/net/e1000.h
> @@ -1134,6 +1134,11 @@ struct e1000_hw {
> boolinitialize_hw_bits_disable;
> e1000_smart_speed   smart_speed;
> e1000_dsp_configdsp_config_state;
> +
> +   struct e1000_rx_desc *rx_base;
> +   unsigned char *rx_packet;
> +   int rx_tail;
> +   int rx_last;
>  };
>
>  #define E1000_EEPROM_SWDPIN0   0x0001  /* SWDPIN 0 EEPROM Value */
> --
> 2.39.2
>


-- 
greets
--
Christian Gmeiner, MSc

https://christian-gmeiner.info/privacypolicy