Skip to contentSkip to conversation

staging/octeon: Mark Ethernet driver as BROKEN

PATCH In review Guenter Roeck 3 participants 7 messages Dec 2, 2019 – Feb 4, 2020 Staging

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 staging/octeon: Mark Ethernet driver as BROKEN

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Unfortunately, one can not just revert the primary offending commit, as doing so results in secondary errors. This is made worse by the fact that the "removed" typedefs still exist and are used widely outside the staging directory, making the entire set of "remove typedef" changes pointless and wrong.

Reflect reality and mark the driver as BROKEN.

Fixes ef1fe6b7369a ("staging: octeon: remove typedef declaration for cvmx_wqe")
Fixes 73aef0c9d2c6 ("staging: octeon: remove typedef declaration for cvmx_helper_link_info")
Cc Wambui Karuga <[email protected]>
Cc Julia Lawall <[email protected]>
---
 drivers/staging/octeon/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/staging/octeon/Kconfig +1 -0
index 5319909eb2f6..e7f4ddcc1361 100644
@@ -3,6 +3,7 @@ config OCTEON_ETHERNET
 	tristate "Cavium Networks Octeon Ethernet support"
 	depends on CAVIUM_OCTEON_SOC || COMPILE_TEST
 	depends on NETDEVICES
+	depends on BROKEN
 	select PHYLIB
 	select MDIO_OCTEON
 	help
Signature · 1 line
2.17.1
#1 lore raw
Quoted 11 lines · Guenter Roeck

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Unfortunately, one can not just revert the primary offending commit, as doing so results in secondary errors. This is made worse by the fact that the "removed" typedefs still exist and are used widely outside the staging directory, making the entire set of "remove typedef" changes pointless and wrong.

Ugh, sorry about that.

Quoted 1 line

Reflect reality and mark the driver as BROKEN.

Should I just delete this thing? No one seems to be using it and there is no move to get it out of staging at all.

Will anyone actually miss it? It can always come back of someone does...

thanks,

greg k-h

#2 lore raw
Quoted 15 lines · Greg Kroah-Hartman
Quoted 11 lines · Guenter Roeck

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Unfortunately, one can not just revert the primary offending commit, as doing so results in secondary errors. This is made worse by the fact that the "removed" typedefs still exist and are used widely outside the staging directory, making the entire set of "remove typedef" changes pointless and wrong.

Ugh, sorry about that.

Quoted 8 lines
Quoted 1 line

Reflect reality and mark the driver as BROKEN.

Should I just delete this thing? No one seems to be using it and there is no move to get it out of staging at all.

Will anyone actually miss it? It can always come back of someone does...

All it does is causing trouble and misguided attempts to clean it up. If anything, the whole thing goes into the wrong direction (declare a complete set of dummy functions just to be able to build the driver with COMPILE_TEST ? Seriously ?).

I second the motion to drop it. This has been in staging for 10 years. Don't we have some kind of time limit for code in staging ? If not, should we ? If anyone really needs it, that person or group should really invest the time to get it out of staging for good.

Guenter

#3 lore raw
Quoted 30 lines · Guenter Roeck
Quoted 15 lines · Greg Kroah-Hartman
Quoted 11 lines · Guenter Roeck

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Unfortunately, one can not just revert the primary offending commit, as doing so results in secondary errors. This is made worse by the fact that the "removed" typedefs still exist and are used widely outside the staging directory, making the entire set of "remove typedef" changes pointless and wrong.

Ugh, sorry about that.

Quoted 8 lines
Quoted 1 line

Reflect reality and mark the driver as BROKEN.

Should I just delete this thing? No one seems to be using it and there is no move to get it out of staging at all.

Will anyone actually miss it? It can always come back of someone does...

All it does is causing trouble and misguided attempts to clean it up. If anything, the whole thing goes into the wrong direction (declare a complete set of dummy functions just to be able to build the driver with COMPILE_TEST ? Seriously ?).

Quoted 4 lines

I second the motion to drop it. This has been in staging for 10 years. Don't we have some kind of time limit for code in staging ? If not, should we ? If anyone really needs it, that person or group should really invest the time to get it out of staging for good.

10 years? Ugh, yes, it's time to drop the thing, I'll do so after -rc1 is out.

thanks,

greg k-h

#4 lore raw

Hi Greg & All,

Quoted 47 lines · Greg Kroah-Hartman
Quoted 40 lines · Guenter Roeck
Quoted 22 lines · Greg Kroah-Hartman
Quoted 10 lines · Guenter Roeck

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from
incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Quoted 8 lines

Unfortunately, one can not just revert the primary offending commit, as doing so
results in secondary errors. This is made worse by the fact that the "removed"
typedefs still exist and are used widely outside the staging directory,
making the entire set of "remove typedef" changes pointless and wrong.

Ugh, sorry about that.

Quoted 9 lines
Quoted 1 line

Reflect reality and mark the driver as BROKEN.

Should I just delete this thing? No one seems to be using it and there
is no move to get it out of staging at all.

Will anyone actually miss it? It can always come back of someone does...

All it does is causing trouble and misguided attempts to clean it up.
If anything, the whole thing goes into the wrong direction (declare a
complete set of dummy functions just to be able to build the driver with COMPILE_TEST ? Seriously ?).

Quoted 5 lines

I second the motion to drop it. This has been in staging for 10 years.
Don't we have some kind of time limit for code in staging ? If not, should we ? If anyone really needs it, that person or group should really invest the time to get it out of staging for good.

Quoted 4 lines

10 years? Ugh, yes, it's time to drop the thing, I'll do so after -rc1
is out.

As a long suffering Cavium MIPs customer could I request that this isn't dropped. I'll get someone here to take a look at fixing the build issues.

Given our platform isn't upstream I'm not sure that we'll be able to meet the criteria for getting it out of staging.

#5 lore raw
Quoted 55 lines · Chris Packham

Hi Greg & All,

Quoted 47 lines · Greg Kroah-Hartman
Quoted 40 lines · Guenter Roeck
Quoted 22 lines · Greg Kroah-Hartman
Quoted 10 lines · Guenter Roeck

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from
incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Quoted 8 lines

Unfortunately, one can not just revert the primary offending commit, as doing so
results in secondary errors. This is made worse by the fact that the "removed"
typedefs still exist and are used widely outside the staging directory,
making the entire set of "remove typedef" changes pointless and wrong.

Ugh, sorry about that.

Quoted 9 lines
Quoted 1 line

Reflect reality and mark the driver as BROKEN.

Should I just delete this thing? No one seems to be using it and there
is no move to get it out of staging at all.

Will anyone actually miss it? It can always come back of someone does...

All it does is causing trouble and misguided attempts to clean it up.
If anything, the whole thing goes into the wrong direction (declare a
complete set of dummy functions just to be able to build the driver with COMPILE_TEST ? Seriously ?).

Quoted 5 lines

I second the motion to drop it. This has been in staging for 10 years.
Don't we have some kind of time limit for code in staging ? If not, should we ? If anyone really needs it, that person or group should really invest the time to get it out of staging for good.

Quoted 4 lines

10 years? Ugh, yes, it's time to drop the thing, I'll do so after -rc1
is out.

Quoted 7 lines

As a long suffering Cavium MIPs customer could I request that this isn't dropped. I'll get someone here to take a look at fixing the build issues.

Given our platform isn't upstream I'm not sure that we'll be able to meet the criteria for getting it out of staging.

Can't you push this onto Cavium as you are paying them for hardware and support?

thanks,

greg k-h

#6 lore raw
Quoted 60 lines · On Tue, 2020-02-04 at 07:36 +0000, [email protected] wrote:
Quoted 59 lines · Chris Packham

Hi Greg & All,

Quoted 47 lines · Greg Kroah-Hartman
Quoted 40 lines · Guenter Roeck
Quoted 22 lines · Greg Kroah-Hartman
Quoted 10 lines · Guenter Roeck

The code doesn't compile due to incompatible pointer errors such as

drivers/staging/octeon/ethernet-tx.c:649:50: error:
passing argument 1 of 'cvmx_wqe_get_grp' from
incompatible pointer type

This is due to mixing, for example, cvmx_wqe_t with 'struct cvmx_wqe'.

Quoted 8 lines

Unfortunately, one can not just revert the primary offending commit, as doing so
results in secondary errors. This is made worse by the fact that the "removed"
typedefs still exist and are used widely outside the staging directory,
making the entire set of "remove typedef" changes pointless and wrong.

Ugh, sorry about that.

Quoted 9 lines
Quoted 1 line

Reflect reality and mark the driver as BROKEN.

Should I just delete this thing? No one seems to be using it and there
is no move to get it out of staging at all.

Will anyone actually miss it? It can always come back of someone does...

All it does is causing trouble and misguided attempts to clean it up.
If anything, the whole thing goes into the wrong direction (declare a
complete set of dummy functions just to be able to build the driver with COMPILE_TEST ? Seriously ?).

Quoted 5 lines

I second the motion to drop it. This has been in staging for 10 years.
Don't we have some kind of time limit for code in staging ? If not, should we ? If anyone really needs it, that person or group should really invest the time to get it out of staging for good.

Quoted 4 lines

10 years? Ugh, yes, it's time to drop the thing, I'll do so after -rc1
is out.

As a long suffering Cavium MIPs customer could I request that this isn't dropped. I'll get someone here to take a look at fixing the build issues.

Quoted 7 lines
Quoted 3 lines

Given our platform isn't upstream I'm not sure that we'll be able to meet the criteria for getting it out of staging.

Can't you push this onto Cavium as you are paying them for hardware and support?

I'll certainly raise this with any and all contacts I have with Cavium/Marvell.