ocfs2/dlm: Move BITS_TO_BYTES() to bitops.h for wider use
Subsystems (5)
PATCH v2 1/1 ocfs2/dlm: Move BITS_TO_BYTES() to bitops.h for wider use
There are users already and will be more of BITS_TO_BYTES() macro. Move it to bitops.h for wider use.
In the case of ocfs2 the replacement is identical.
As for bnx2x, there are two places where floor version is used.
In the first case to calculate the amount of structures that can fit one memory page. In this case obviously the ceiling variant is correct and original code might have a potential bug, if amount of bits % 8 is not 0. In the second case the macro is used to calculate bytes transmitted in one microsecond. This will work for all speeds which is multiply of 1Gbps without any change, for the rest new code will give ceiling value, for instance 100Mbps will give 13 bytes, while old code gives 12 bytes and the arithmetically correct one is 12.5 bytes. Further the value is used to setup timer threshold which in any case has its own margins due to certain resolution. I don't see here an issue with slightly shifting thresholds for low speed connections, the card is supposed to utilize highest available rate, which is usually 10Gbps.
---
- described bnx2x cases in the commit message
- appended Rb (for ocfs2)
drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h | 1 - fs/ocfs2/dlm/dlmcommon.h | 4 ---- include/linux/bitops.h | 1 + 3 files changed, 1 insertion(+), 5 deletions(-)
drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h +0 -1
@@ -296,7 +296,6 @@ static inline void bnx2x_dcb_config_qm(struct bnx2x *bp, enum cos_mode mode,* possible, the driver should only write the valid vnics into the internal * ram according to the appropriate port mode. */ -#define BITS_TO_BYTES (x) ((x)/8)/* CMNG constants, as derived from system spec calculations */
fs/ocfs2/dlm/dlmcommon.h +0 -4
@@ -688,10 +688,6 @@ struct dlm_begin_reco__be32 pad2; }; - -#define BITS_PER_BYTE 8 -#define BITS_TO_BYTES (bits) (((bits)+BITS_PER_BYTE-1)/BITS_PER_BYTE) -struct dlm_query_join_request {u8 node_idx;
include/linux/bitops.h +1 -0
@@ -5,6 +5,7 @@#include <linux/bits.h> #define BITS_PER_TYPE (type) (sizeof(type) * BITS_PER_BYTE) +#define BITS_TO_BYTES (nr) DIV_ROUND_UP(nr, BITS_PER_BYTE)#define BITS_TO_LONGS (nr) DIV_ROUND_UP(nr, BITS_PER_TYPE(long))extern unsigned int __sw_hweight8 (unsigned int w);
Signature · 1 line
2.23.0.rc1
Quoted 5 lines · Andy Shevchenko
There are users already and will be more of
BITS_TO_BYTES()macro. Move it tobitops.hfor wider use.In the case of ocfs2 the replacement is identical.
Quoted 12 lines
As for bnx2x, there are two places where floor version is used.
In the first case to calculate the amount of structures that can fit one memory page. In this case obviously the ceiling variant is correct and original code might have a potential bug, if amount of bits % 8 is not 0. In the second case the macro is used to calculate bytes transmitted in one microsecond. This will work for all speeds which is multiply of 1Gbps without any change, for the rest new code will give ceiling value, for instance 100Mbps will give 13 bytes, while old code gives 12 bytes and the arithmetically correct one is 12.5 bytes. Further the value is used to setup timer threshold which in any case has its own margins due to certain resolution. I don't see here an issue with slightly shifting thresholds for low speed connections, the card is supposed to utilize highest available rate, which is usually 10Gbps.
Anybody to comment on bnx2 change?
Can we survive with this applied?
Quoted 54 lines
Reviewed-by Joseph Qi <[email protected]>Signed-off-by Andy Shevchenko <[email protected]>---
- described bnx2x cases in the commit message
- appended Rb (for ocfs2)drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h | 1 - fs/ocfs2/dlm/dlmcommon.h | 4 ---- include/linux/bitops.h | 1 + 3 files changed, 1 insertion(+), 5 deletions(-)drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h +0
-1@@ -296,7 +296,6 @@ static inline void bnx2x_dcb_config_qm(struct bnx2x *bp, enum cos_mode mode,* possible, the driver should only write the valid vnics into the internal * ram according to the appropriate port mode. */ -#define BITS_TO_BYTES (x) ((x)/8)/* CMNG constants, as derived from system spec calculations */ fs/ocfs2/dlm/dlmcommon.h +0
-4@@ -688,10 +688,6 @@ struct dlm_begin_reco__be32 pad2; }; - -#define BITS_PER_BYTE 8 -#define BITS_TO_BYTES (bits) (((bits)+BITS_PER_BYTE-1)/BITS_PER_BYTE) -struct dlm_query_join_request {u8 node_idx;include/linux/bitops.h +1
-0@@ -5,6 +5,7 @@#include <linux/bits.h> #define BITS_PER_TYPE (type) (sizeof(type) * BITS_PER_BYTE) +#define BITS_TO_BYTES (nr) DIV_ROUND_UP(nr, BITS_PER_BYTE)#define BITS_TO_LONGS (nr) DIV_ROUND_UP(nr, BITS_PER_TYPE(long))extern unsigned int __sw_hweight8 (unsigned int w);Signature · 1 line
2.23.0.rc1
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 88 lines
-----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Andy Shevchenko
Sent: Monday, December 9, 2019 10:14 PM
To: Mark Fasheh <[email protected]>; Joel Becker <[email protected]>; Joseph Qi <[email protected]>; [email protected]; Ariel Elior <[email protected]>; Sudarsana Reddy Kalluru <[email protected]>; GR-everest-linux-l2 <GR-everest-linux- [email protected]>; David S. Miller <[email protected]>; [email protected]; Colin Ian King <[email protected]>
Subject: Re: [PATCH v2] ocfs2/dlm: MoveBITS_TO_BYTES()tobitops.hfor wider useQuoted 5 lines · Andy Shevchenko
There are users already and will be more of
BITS_TO_BYTES()macro. Move it tobitops.hfor wider use.In the case of ocfs2 the replacement is identical.
Quoted 12 lines
As for bnx2x, there are two places where floor version is used. In the first case to calculate the amount of structures that can fit one memory page. In this case obviously the ceiling variant is correct and original code might have a potential bug, if amount of bits % 8 is not 0. In the second case the macro is used to calculate bytes transmitted in one microsecond. This will work for all speeds which is multiply of 1Gbps without any change, for the rest new code will give ceiling value, for instance 100Mbps will give 13 bytes, while old code gives 12 bytes and the arithmetically correct one is 12.5 bytes. Further the value is used to setup timer threshold which in any case has its own margins due to certain resolution. I don't see here an issue with slightly shifting thresholds for low speed connections, the card is supposed
to utilize highest available rate, which is usually 10Gbps.
Anybody to comment on bnx2 change?
Can we survive with this applied?drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h static inline void bnx2x_dcb_config_qm(struct bnx2x
Quoted 18 lines
Reviewed-by Joseph Qi <[email protected]>Signed-off-by Andy Shevchenko <[email protected]>---
- described bnx2x cases in the commit message
- appended Rb (for ocfs2)drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h | 1 - fs/ocfs2/dlm/dlmcommon.h | 4 ---- include/linux/bitops.h | 1 + 3 files changed, 1 insertion(+), 5 deletions(-)drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h +0
-0@@ -296,7 +296,6 @@ static inline void bnx2x_dcb_config_qm(struct bnx2x*bp, enum
cos_modemode,Quoted 33 lines
drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h +0
-1* possible, the driver should only write the valid vnics into the internal * ram according to the appropriate port mode. */ -#define BITS_TO_BYTES (x) ((x)/8)/* CMNG constants, as derived from system spec calculations */ fs/ocfs2/dlm/dlmcommon.h index renamed from fs/ocfs2/dlm/dlmcommon.h +0
-0
aaf24548b02a..0463dce65bb2100644fs/ocfs2/dlm/dlmcommon.h +0
-4@@ -688,10 +688,6 @@ struct dlm_begin_reco__be32 pad2; }; - -#define BITS_PER_BYTE 8 -#define BITS_TO_BYTES (bits) (((bits)+BITS_PER_BYTE-1)/BITS_PER_BYTE) -struct dlm_query_join_request {u8 node_idx;include/linux/bitops.h index renamed from include/linux/bitops.h +0
-0
cf074bce3eb3..79d80f5ddf7b100644include/linux/bitops.h +1
-0@@ -5,6 +5,7 @@#include <linux/bits.h> #define BITS_PER_TYPE (type) (sizeof(type) * BITS_PER_BYTE) +#define BITS_TO_BYTES (nr) DIV_ROUND_UP(nr, BITS_PER_BYTE)#define BITS_TO_LONGS (nr) DIV_ROUND_UP(nr, BITS_PER_TYPE(long))
Quoted 9 lines
Quoted 4 lines
include/linux/bitops.h +0
-1extern unsigned int __sw_hweight8 (unsigned int w); --2.23.0.rc1
--
With Best Regards,
Andy Shevchenko
Thanks for the changes.
Joseph Qi
maintainerAdd akpm as well.
Quoted 102 lines · Sudarsana Reddy Kalluru
Quoted 88 lines
-----Original Message-----
From: netdev-owner at vger.kernel.org <[email protected]> On Behalf Of Andy Shevchenko
Sent: Monday, December 9, 2019 10:14 PM
To: Mark Fasheh <[email protected]>; Joel Becker <[email protected]>; Joseph Qi <[email protected]>; ocfs2-devel at oss.oracle.com; Ariel Elior <[email protected]>; Sudarsana Reddy Kalluru <[email protected]>; GR-everest-linux-l2 <GR-everest-linux-
l2 at marvell.com>; David S. Miller <[email protected]>;
netdev at vger.kernel.org; Colin Ian King <[email protected]>
Subject: Re: [PATCH v2] ocfs2/dlm: MoveBITS_TO_BYTES()tobitops.hfor wider useQuoted 5 lines · Andy Shevchenko
There are users already and will be more of
BITS_TO_BYTES()macro. Move it tobitops.hfor wider use.In the case of ocfs2 the replacement is identical.
Quoted 12 lines
As for bnx2x, there are two places where floor version is used. In the first case to calculate the amount of structures that can fit one memory page. In this case obviously the ceiling variant is correct and original code might have a potential bug, if amount of bits % 8 is not 0. In the second case the macro is used to calculate bytes transmitted in one microsecond. This will work for all speeds which is multiply of 1Gbps without any change, for the rest new code will give ceiling value, for instance 100Mbps will give 13 bytes, while old code gives 12 bytes and the arithmetically correct one is 12.5 bytes. Further the value is used to setup timer threshold which in any case has its own margins due to certain resolution. I don't see here an issue with slightly shifting thresholds for low speed connections, the card is supposed
to utilize highest available rate, which is usually 10Gbps.
Anybody to comment on bnx2 change?
Can we survive with this applied?drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h static inline void bnx2x_dcb_config_qm(struct bnx2x
Quoted 18 lines
Reviewed-by Joseph Qi <[email protected]>Signed-off-by Andy Shevchenko <[email protected]>---
- described bnx2x cases in the commit message
- appended Rb (for ocfs2)drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h | 1 - fs/ocfs2/dlm/dlmcommon.h | 4 ---- include/linux/bitops.h | 1 + 3 files changed, 1 insertion(+), 5 deletions(-)drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h +0
-0@@ -296,7 +296,6 @@ static inline void bnx2x_dcb_config_qm(struct bnx2x*bp, enum
cos_modemode,Quoted 33 lines
drivers/net/ethernet/broadcom/bnx2x/bnx2x_init.h +0
-1* possible, the driver should only write the valid vnics into the internal * ram according to the appropriate port mode. */ -#define BITS_TO_BYTES (x) ((x)/8)/* CMNG constants, as derived from system spec calculations */ fs/ocfs2/dlm/dlmcommon.h index renamed from fs/ocfs2/dlm/dlmcommon.h +0
-0
aaf24548b02a..0463dce65bb2100644fs/ocfs2/dlm/dlmcommon.h +0
-4@@ -688,10 +688,6 @@ struct dlm_begin_reco__be32 pad2; }; - -#define BITS_PER_BYTE 8 -#define BITS_TO_BYTES (bits) (((bits)+BITS_PER_BYTE-1)/BITS_PER_BYTE) -struct dlm_query_join_request {u8 node_idx;include/linux/bitops.h index renamed from include/linux/bitops.h +0
-0
cf074bce3eb3..79d80f5ddf7b100644include/linux/bitops.h +1
-0@@ -5,6 +5,7 @@#include <linux/bits.h> #define BITS_PER_TYPE (type) (sizeof(type) * BITS_PER_BYTE) +#define BITS_TO_BYTES (nr) DIV_ROUND_UP(nr, BITS_PER_BYTE)#define BITS_TO_LONGS (nr) DIV_ROUND_UP(nr, BITS_PER_TYPE(long))Quoted 9 lines
Quoted 4 lines
include/linux/bitops.h +0
-1extern unsigned int __sw_hweight8 (unsigned int w); --2.23.0.rc1
--
With Best Regards,
Andy ShevchenkoThanks for the changes.