Re: [PATCH net-next 2/2] bonding: combine netlink and console error messages

From: Jonathan Toppins
Date: Sat Aug 07 2021 - 17:54:44 EST


On 8/6/21 11:52 PM, Joe Perches wrote:
On Fri, 2021-08-06 at 23:30 -0400, Jonathan Toppins wrote:
There seems to be no reason to have different error messages between
netlink and printk. It also cleans up the function slightly.
[]
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
[]
+#define BOND_NL_ERR(bond_dev, extack, errmsg) do { \
+ NL_SET_ERR_MSG(extack, errmsg); \
+ netdev_err(bond_dev, "Error: " errmsg "\n"); \
+} while (0)
+
+#define SLAVE_NL_ERR(bond_dev, slave_dev, extack, errmsg) do { \
+ NL_SET_ERR_MSG(extack, errmsg); \
+ slave_err(bond_dev, slave_dev, "Error: " errmsg "\n"); \
+} while (0)

If you are doing this, it's probably smaller object code to use
"%s", errmsg
as the errmsg string can be reused

#define BOND_NL_ERR(bond_dev, extack, errmsg) \
do { \
NL_SET_ERR_MSG(extack, errmsg); \
netdev_err(bond_dev, "Error: %s\n", errmsg); \
} while (0)

#define SLAVE_NL_ERR(bond_dev, slave_dev, extack, errmsg) \
do { \
NL_SET_ERR_MSG(extack, errmsg); \
slave_err(bond_dev, slave_dev, "Error: %s\n", errmsg); \
} while (0)



I like the thought and would agree if not for how NL_SET_ERR_MSG is coded. Unfortunately it does not appear as though doing the above change actually generates smaller object code. Maybe I have incorrectly interpreted something?

$ git show
commit 6bb346263b4e9d008744b45af5105df309c35c1a (HEAD -> upstream-bonding-cleanup)
Author: Jonathan Toppins <jtoppins@xxxxxxxxxx>
Date: Sat Aug 7 17:34:58 2021 -0400

object code optimization

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 46b95175690b..e2903ae7cdab 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -1714,12 +1714,12 @@ void bond_lower_state_changed(struct slave *slave)

#define BOND_NL_ERR(bond_dev, extack, errmsg) do { \
NL_SET_ERR_MSG(extack, errmsg); \
- netdev_err(bond_dev, "Error: " errmsg "\n"); \
+ netdev_err(bond_dev, "Error: %s\n", errmsg); \
} while (0)

#define SLAVE_NL_ERR(bond_dev, slave_dev, extack, errmsg) do { \
NL_SET_ERR_MSG(extack, errmsg); \
- slave_err(bond_dev, slave_dev, "Error: " errmsg "\n"); \
+ slave_err(bond_dev, slave_dev, "Error: %s\n", errmsg); \
} while (0)

/* enslave device <slave> to bond device <master> */
$ git log --oneline -3
6bb346263b4e (HEAD -> upstream-bonding-cleanup) object code optimization
a36c7639a139 bonding: combine netlink and console error messages
88916c847e85 bonding: remove extraneous definitions from bonding.h
jtoppins@jtoppins:~/projects/linux-rhel7$ git rebase -i --exec "make drivers/net/bonding/bond_main.o; ls -l drivers/net/bonding/bond_main.o" HEAD^^
hint: Waiting for your editor to close the file... Error detected while processing /home/jtoppins/.vim/bundle/cscope_macros.vim/plugin/cscope_macros.vim:
line 42:
E568: duplicate cscope database not added
Press ENTER or type command to continue
Executing: make menuconfig


*** End of the configuration.
*** Execute 'make' to start the build or try 'make help'.

Executing: make drivers/net/bonding/bond_main.o; ls -l drivers/net/bonding/bond_main.o
CALL scripts/checksyscalls.sh
CALL scripts/atomic/check-atomics.sh
DESCEND objtool
DESCEND bpf/resolve_btfids
CC [M] drivers/net/bonding/bond_main.o
-rw-r--r--. 1 jtoppins jtoppins 1777896 Aug 7 17:37 drivers/net/bonding/bond_main.o
Executing: make drivers/net/bonding/bond_main.o; ls -l drivers/net/bonding/bond_main.o
CALL scripts/checksyscalls.sh
CALL scripts/atomic/check-atomics.sh
DESCEND objtool
DESCEND bpf/resolve_btfids
CC [M] drivers/net/bonding/bond_main.o
-rw-r--r--. 1 jtoppins jtoppins 1778320 Aug 7 17:37 drivers/net/bonding/bond_main.o
Successfully rebased and updated refs/heads/upstream-bonding-cleanup.

It appears the change increases bond_main.o by 424 (1778320-1777896) bytes.