Skip to contentSkip to conversation

igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

PATCH Superseded Sasha Neftin 5 participants 10 messages Aug 23, 2018 – Aug 27, 2018

Subsystems (2)

v6 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#0 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

This patch adds the beginning framework onto which I am going to add the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

Sasha Neftin (v5):
no changes

Sasha Neftin (v6):
no changes

---
 drivers/net/ethernet/intel/Kconfig        |  16 +++
 drivers/net/ethernet/intel/Makefile       |   1 +
 drivers/net/ethernet/intel/igc/Makefile   |  10 ++
 drivers/net/ethernet/intel/igc/igc.h      |  29 +++++
 drivers/net/ethernet/intel/igc/igc_hw.h   |  10 ++
 drivers/net/ethernet/intel/igc/igc_main.c | 146 ++++++++++++++++++++++
 6 files changed, 212 insertions(+)
 create mode 100644 drivers/net/ethernet/intel/igc/Makefile
 create mode 100644 drivers/net/ethernet/intel/igc/igc.h
 create mode 100644 drivers/net/ethernet/intel/igc/igc_hw.h
 create mode 100644 drivers/net/ethernet/intel/igc/igc_main.c
drivers/net/ethernet/intel/Kconfig +16 -0
index 1ab613eb5796..fcb676398022 100644
@@ -280,4 +280,20 @@ config FM10K
 	  To compile this driver as a module, choose M here. The module
 	  will be called fm10k.  MSI-X interrupt support is required
 
+config IGC
+	tristate "Intel(R) Ethernet Controller I225-LM/I225-V support"
+	default n
+	depends on PCI
+	---help---
+	  This driver supports Intel(R) Ethernet Controller I225-LM/I225-V
+	  family of adapters.
+
+	  For more information on how to identify your adapter, go
+	  to the Adapter & Driver ID Guide that can be located at:
+
+	  <http://support.intel.com>
+
+	  To compile this driver as a module, choose M here. The module
+	  will be called igc.
+
 endif # NET_VENDOR_INTEL
drivers/net/ethernet/intel/Makefile +1 -0
index 807a4f8c7e4e..18f4aac68bc0 100644
@@ -7,6 +7,7 @@ obj-$(CONFIG_E100) += e100.o
 obj-$(CONFIG_E1000) += e1000/
 obj-$(CONFIG_E1000E) += e1000e/
 obj-$(CONFIG_IGB) += igb/
+obj-$(CONFIG_IGC) += igc/
 obj-$(CONFIG_IGBVF) += igbvf/
 obj-$(CONFIG_IXGBE) += ixgbe/
 obj-$(CONFIG_IXGBEVF) += ixgbevf/
drivers/net/ethernet/intel/igc/Makefile new file +10 -0
new file mode 100644
index 000000000000..3d13b015d401
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c)  2018 Intel Corporation
+
+#
+# Intel(R) I225-LM/I225-V 2.5G Ethernet Controller
+#
+
+obj-$(CONFIG_IGC) += igc.o
+
+igc-objs := igc_main.o
drivers/net/ethernet/intel/igc/igc.h new file +29 -0
new file mode 100644
index 000000000000..afe595cfcf63
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c)  2018 Intel Corporation */
+
+#ifndef _IGC_H_
+#define _IGC_H_
+
+#include <linux/kobject.h>
+
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/vmalloc.h>
+
+#include <linux/ethtool.h>
+
+#include <linux/sctp.h>
+
+#define IGC_ERR(args...) pr_err("igc: " args)
+
+#define PFX "igc: "
+
+#include <linux/timecounter.h>
+#include <linux/net_tstamp.h>
+#include <linux/ptp_clock_kernel.h>
+
+/* main */
+extern char igc_driver_name[];
+extern char igc_driver_version[];
+
+#endif /* _IGC_H_ */
drivers/net/ethernet/intel/igc/igc_hw.h new file +10 -0
new file mode 100644
index 000000000000..aa68b4516700
@@ -0,0 +1,10 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c)  2018 Intel Corporation */
+
+#ifndef _IGC_HW_H_
+#define _IGC_HW_H_
+
+#define IGC_DEV_ID_I225_LM			0x15F2
+#define IGC_DEV_ID_I225_V			0x15F3
+
+#endif /* _IGC_HW_H_ */
drivers/net/ethernet/intel/igc/igc_main.c new file +146 -0

146 lines not on this page · raw message

Signature · 1 line
2.18.0
#1 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 3 lines · Sasha Neftin

This patch adds the beginning framework onto which I am going to add the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

I'm curious. The code looks very much like the igb code. Wouldn't it be simpler and less error prone to add these new NICs to igb, rather than creating a new driver by (mostly) copying the code from igb?

What's the idea behind this?

Quoted 11 lines

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

In theory this is neither necessary, nor does it actually help. By keeping the e1000 prefixes it's much easier to compare to the other, already existing drivers.

Quoted 5 lines

Sasha Neftin (v5):
no changes

Sasha Neftin (v6):
no changes

Is that an oversight? If there are no changes in v5 and v6, why do we have them?!?

Corinna
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/a312b52b/attachment.asc>

#2 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 6 lines · Corinna Vinschen
Quoted 1 line · Sasha Neftin

This patch adds the beginning framework onto which I am going to

add

Quoted 2 lines

the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

Quoted 6 lines

I'm curious. The code looks very much like the igb code. Wouldn't it
be simpler and less error prone to add these new NICs to igb, rather than creating a new driver by (mostly) copying the code from igb?

What's the idea behind this?

This new driver is for a new upcoming client part, not server (which is the igb driver). Yes, the way that this new client part does resemble how the igb driver operates, but this new client part will not have many of the advance features (like SRIOV) that igb uses. We also want to keep the igb driver strictly for the 1GbE server devices.

Quoted 17 lines
Quoted 5 lines

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Quoted 7 lines

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

In theory this is neither necessary, nor does it actually help. By keeping the e1000 prefixes it's much easier to compare to the other, already existing drivers.

Quoted 8 lines
Quoted 5 lines

Sasha Neftin (v5):
no changes

Sasha Neftin (v6):
no changes

Is that an oversight? If there are no changes in v5 and v6, why do we have them?!?

Sasha is just noting that this particular patch did not change in v6,
but other patches did have changes in this new version of the series.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/b52fee5c/attachment.asc>

#3 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 12 lines · Jeff Kirsher
Quoted 11 lines · Corinna Vinschen
Quoted 1 line · Sasha Neftin

This patch adds the beginning framework onto which I am going to

add

Quoted 2 lines

the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

I'm curious. The code looks very much like the igb code. Wouldn't it
be simpler and less error prone to add these new NICs to igb, rather than creating a new driver by (mostly) copying the code from igb?

Quoted 7 lines
Quoted 1 line

What's the idea behind this?

This new driver is for a new upcoming client part, not server (which is the igb driver). Yes, the way that this new client part does resemble how the igb driver operates, but this new client part will not have many of the advance features (like SRIOV) that igb uses. We also want to keep the igb driver strictly for the 1GbE server devices.

Ok, thanks, but I'm still a bit puzzled.

What about E1000_DEV_ID_I354_BACKPLANE_2_5GBPS? The later supposedly already provides 2.5G within igb. Is that the server part compared to igc? Otherwise, what's the equivalent server part providing 2.5G?

Quoted 19 lines
Quoted 19 lines
Quoted 11 lines

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

In theory this is neither necessary, nor does it actually help. By keeping the e1000 prefixes it's much easier to compare to the other, already existing drivers.

Quoted 3 lines

Sasha Neftin (v5):
no changes

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

Sasha Neftin (v6):
no changes

Is that an oversight? If there are no changes in v5 and v6, why do we have them?!?

Sasha is just noting that this particular patch did not change in v6, but other patches did have changes in this new version of the series.

Oh, right, thanks for the clarification.

Corinna
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/8490f85d/attachment.asc>

#4 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 24 lines · Corinna Vinschen
Quoted 3 lines · Jeff Kirsher
Quoted 2 lines · Corinna Vinschen
Quoted 1 line · Sasha Neftin

This patch adds the beginning framework onto which I am going

to

Quoted 5 lines
Quoted 5 lines

add

Quoted 2 lines

the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

I'm curious. The code looks very much like the igb code.

Wouldn't

Quoted 2 lines
Quoted 2 lines

it
be simpler and less error prone to add these new NICs to igb,

rather

Quoted 5 lines
Quoted 3 lines

than creating a new driver by (mostly) copying the code from igb?

What's the idea behind this?

This new driver is for a new upcoming client part, not server

(which is

Quoted 1 line

the igb driver). Yes, the way that this new client part does

resemble

Quoted 2 lines

how the igb driver operates, but this new client part will not have many of the advance features (like SRIOV) that igb uses. We also

Quoted 8 lines

want

Quoted 1 line

to keep the igb driver strictly for the 1GbE server devices.

Ok, thanks, but I'm still a bit puzzled.

What about E1000_DEV_ID_I354_BACKPLANE_2_5GBPS? The later supposedly already provides 2.5G within igb. Is that the server part compared to igc? Otherwise, what's the equivalent server part providing 2.5G?

Yes, one of the last server devices we developed back in 2013 was I354 and that device did support 2.5 GbE. Again, this is a server part, which supports more advanced features that client parts do not.

There is not a 1:1 on server parts and client parts, sometimes there can be a server part that is similar to a client part or vice versa.

In this case, this new client part will be the next generation of client parts. So these will be essentially the new "e1000e" (our other client driver), but since this next generation of client parts will have MAC/PHY communications that are more similar to how some of our server parts did, that is why portions of the driver will resemble igb. But feature wise the igc driver will be more like e1000e than igb.

Hope that helps.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/9830e4dc/attachment-0001.asc>

#5 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 41 lines · Jeff Kirsher
Quoted 24 lines · Corinna Vinschen
Quoted 3 lines · Jeff Kirsher
Quoted 2 lines · Corinna Vinschen
Quoted 1 line · Sasha Neftin

This patch adds the beginning framework onto which I am going

to

Quoted 5 lines
Quoted 5 lines

add

Quoted 2 lines

the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

I'm curious. The code looks very much like the igb code.

Wouldn't

Quoted 2 lines
Quoted 2 lines

it
be simpler and less error prone to add these new NICs to igb,

rather

Quoted 5 lines
Quoted 3 lines

than creating a new driver by (mostly) copying the code from igb?

What's the idea behind this?

This new driver is for a new upcoming client part, not server

(which is

Quoted 1 line

the igb driver). Yes, the way that this new client part does

resemble

Quoted 2 lines

how the igb driver operates, but this new client part will not have many of the advance features (like SRIOV) that igb uses. We also

Quoted 8 lines

want

Quoted 1 line

to keep the igb driver strictly for the 1GbE server devices.

Ok, thanks, but I'm still a bit puzzled.

What about E1000_DEV_ID_I354_BACKPLANE_2_5GBPS? The later supposedly already provides 2.5G within igb. Is that the server part compared to igc? Otherwise, what's the equivalent server part providing 2.5G?

Yes, one of the last server devices we developed back in 2013 was I354 and that device did support 2.5 GbE. Again, this is a server part, which supports more advanced features that client parts do not.

There is not a 1:1 on server parts and client parts, sometimes there can be a server part that is similar to a client part or vice versa.

Quoted 8 lines

In this case, this new client part will be the next generation of client parts. So these will be essentially the new "e1000e" (our other client driver), but since this next generation of client parts will have MAC/PHY communications that are more similar to how some of our server parts did, that is why portions of the driver will resemble igb. But feature wise the igc driver will be more like e1000e than igb.

Hope that helps.

Yes, Jeff, this really helps.

But this just reinforces my doubts that changing the E1000 prefix to IGC is a good thing. The comparison of binary values only gets more complicated. In the end we have, for instance, this:

  e1000:  #define E1000_ERR_NVM        1
  e1000e: #define E1000_ERR_NVM        1
  igb:    #define E1000_ERR_NVM        1

but

igc: #define IGC_ERR_NVM 1

This doesn't make sense.

Thanks,
Corinna
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/f2087ca4/attachment.asc>

#6 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 75 lines · Corinna Vinschen
Quoted 5 lines · Jeff Kirsher
Quoted 4 lines · Corinna Vinschen
Quoted 3 lines · Jeff Kirsher
Quoted 2 lines · Corinna Vinschen
Quoted 1 line · Sasha Neftin

This patch adds the beginning framework onto which I am

going

Quoted 3 lines
Quoted 3 lines

to

Quoted 2 lines
Quoted 2 lines

add

Quoted 1 line

the igc driver which supports the Intel(R) I225-LM/I225-V

2.5G

Quoted 8 lines
Quoted 8 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

Ethernet Controller.

I'm curious. The code looks very much like the igb code.

Wouldn't

Quoted 2 lines
Quoted 2 lines

it
be simpler and less error prone to add these new NICs to igb,

rather

Quoted 1 line
Quoted 1 line

than creating a new driver by (mostly) copying the code from

igb?

Quoted 8 lines
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

What's the idea behind this?

This new driver is for a new upcoming client part, not server

(which is

Quoted 1 line

the igb driver). Yes, the way that this new client part does

resemble

Quoted 1 line

how the igb driver operates, but this new client part will not

have

Quoted 1 line
Quoted 1 line
Quoted 1 line

many of the advance features (like SRIOV) that igb uses. We

also

Quoted 6 lines
Quoted 6 lines

want

Quoted 1 line

to keep the igb driver strictly for the 1GbE server devices.

Ok, thanks, but I'm still a bit puzzled.

What about E1000_DEV_ID_I354_BACKPLANE_2_5GBPS? The later

supposedly

Quoted 1 line
Quoted 1 line

already provides 2.5G within igb. Is that the server part

compared

Quoted 1 line
Quoted 1 line

to igc? Otherwise, what's the equivalent server part providing

2.5G?

Quoted 2 lines

Yes, one of the last server devices we developed back in 2013 was

I354

Quoted 4 lines

and that device did support 2.5 GbE. Again, this is a server part, which supports more advanced features that client parts do not.

There is not a 1:1 on server parts and client parts, sometimes

there

Quoted 1 line

can be a server part that is similar to a client part or vice

versa.

Quoted 3 lines

In this case, this new client part will be the next generation of client parts. So these will be essentially the new "e1000e" (our

other

Quoted 2 lines

client driver), but since this next generation of client parts will have MAC/PHY communications that are more similar to how some of

our

Quoted 1 line

server parts did, that is why portions of the driver will resemble

igb.

Quoted 3 lines

But feature wise the igc driver will be more like e1000e than igb.

Hope that helps.

Yes, Jeff, this really helps.

But this just reinforces my doubts that changing the E1000 prefix to IGC is a good thing. The comparison of binary values only gets more complicated. In the end we have, for instance, this:

  e1000:  #define E1000_ERR_NVM        1
  e1000e: #define E1000_ERR_NVM        1
  igb:    #define E1000_ERR_NVM        1
Quoted 5 lines

but

igc: #define IGC_ERR_NVM 1

This doesn't make sense.

When naming functions/macros and defines, I can see pros and cons to both approaches. That particular code you referring to is "common code" shared amongst the various drivers e1000, e1000e and igb. The igc driver did start with this "common code" as a starting point, but this common code will be morphing to handle the future client parts. So we were looking to use the igc driver as the parent of this new common code, much like how e1000 was the "original" or parent driver to e1000e and igb.

It seemed right to break ties with the legacy naming of
functions/macros/defines, especially since this new generation of parts
and devices will have very little in common with the old e1000 devices. -------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/6887dbd2/attachment.asc>

#7 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

In reply to Sasha Neftin
Quoted 16 lines · Sasha Neftin

This patch adds the beginning framework onto which I am going to add the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

Quoted 5 lines

Sasha Neftin (v5):
no changes

Sasha Neftin (v6):
no changes

In all these patches, this version stuff will be removed from the commit message before pushing upstream, right?

Normally they appear after the "---" so as to not be a part of the formal commit message saved in git.

[...]

Quoted 7 lines
drivers/net/ethernet/intel/igc/igc.h new file +1 -0
new file mode 100644
index 000000000000..afe595cfcf63
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */

[...]

Quoted 1 line

+MODULE_LICENSE("GPL");

Since you're using the SPDX "GPL-2.0", you might want to use the MODULE_LICENSE string "GPL v2" to match. since "GPL" means v2 or later. See include/linux/module.h for details.

By the way, this all looks much nicer with the "igc_" prefix - thanks.

sln

#8 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

Quoted 24 lines · Shannon Nelson
Quoted 17 lines · Sasha Neftin

This patch adds the beginning framework onto which I am going to add
the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

Quoted 5 lines

Sasha Neftin (v5):
no changes

Sasha Neftin (v6):
no changes

Quoted 6 lines

In all these patches, this version stuff will be removed from the commit
message before pushing upstream, right?

Normally they appear after the "---" so as to not be a part of the formal commit message saved in git.

Yes, when this eventually gets pushed upstream, the changelog will be
stripped.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20180823/2dd489bb/attachment.asc>

#9 lore raw

PATCH v6 1/11 igc: Add skeletal frame for Intel(R) 2.5G Ethernet Controller support.

In reply to Shannon Nelson
Quoted 44 lines · Shannon Nelson
Quoted 16 lines · Sasha Neftin

This patch adds the beginning framework onto which I am going to add the igc driver which supports the Intel(R) I225-LM/I225-V 2.5G Ethernet Controller.

Sasha Neftin (v2):
update module author, copyright and licence header cosmetic changes

Sasha Neftin (v3):
remove unused header files

Sasha Neftin (v4):
update brand name
fix syntax by input from community
replace e1000_ prefix with igc_ prefix

Quoted 5 lines

Sasha Neftin (v5):
no changes

Sasha Neftin (v6):
no changes

In all these patches, this version stuff will be removed from the commit message before pushing upstream, right?

Normally they appear after the "---" so as to not be a part of the formal commit message saved in git.

[...]

Quoted 8 lines
drivers/net/ethernet/intel/igc/igc.h new file +1 -0
new file mode 100644
index 000000000000..afe595cfcf63
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */

[...]

Quoted 1 line

+MODULE_LICENSE("GPL");

Quoted 8 lines

Since you're using the SPDX "GPL-2.0", you might want to use the MODULE_LICENSE string "GPL v2" to match. since "GPL" means v2 or later. See include/linux/module.h for details.

By the way, this all looks much nicer with the "igc_" prefix - thanks.

sln
Our new drivers, ice, i40e, fm10k used same convention for prefix

naming. We decided keep igc_ prefix.
Thanks,
Sasha