Dear Kim Phillips,
In message 20100719193825.120ebf29.kim.phillips@freescale.com you wrote:
- It would be nice if we could add wildcard support for environment variables; this is needed for variable name auto-completion, but it would also be nice to be able to say "printenv ip*" or "printenv *addr*"
you were right - a grepenv/findenv/'env search' substring implementation on top of this looks to be at least as expensive as a full export operation :/
Expensive in terms of what? Code size? Probably not. It all boils down to running strstr() over all entries...
int var_complete(int argc, char * const argv[], char last_char, int maxv, char *cmdv[]) { +#if 0 /* need to reimplement */
ouch - this is u-boot's most useful feature :)
He. We should probably run a poll for UMUF :-)
It would be good to know boot time overhead the initial import function makes, esp. in terms of number of boot-time accesses to the environment...
I don't see any significant changes on the systems I tested...
Example TQM5200:
MPC5200 at 400 MHz, 16 KiB environment size, ~2 KiB used:
relevant env settings:
nfsargs=setenv bootargs root=/dev/nfs rw nfsroot=${serverip}:${rootpath} addip=setenv bootargs ${bootargs} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}:${netdev}:off panic=1 addcons=setenv bootargs ${bootargs} console=${console},${baudrate} net_nfs=tftp ${kernel_addr_r} ${bootfile}; tftp ${fdt_addr_r} ${fdt_file}; run nfsargs addip addcons; bootm ${kernel_addr_r} - ${fdt_addr_r} bootcmd=run net_nfs
old: U-Boot 2010.03-00167-g6e5fb4e (Apr 27 2010 - 15:37:15) new: U-Boot 2010.06-00208-g7151bce-dirty (Jul 18 2010 - 00:36:48)
old new 0.000 0.000 U-Boot 2010.03-00167-g6e5fb4e (Apr 27 2010 - 15:37:15) 0.000 0.000 0.000 0.000 CPU: MPC5200B v2.2, Core v1.4 at 396 MHz 0.000 0.000 Bus 132 MHz, IPB 132 MHz, PCI 66 MHz 0.040 0.040 Board: TQM5200S (TQ-Components GmbH) 0.040 0.040 on a STK52xx carrier board 0.040 0.040 I2C: 85 kHz, ready 0.040 0.040 DRAM: 64 MB 0.080 0.080 FLASH: 32 MB 0.120 0.130 In: serial 0.120 0.130 Out: serial 0.120 0.130 Err: serial 0.160 0.130 Net: FEC ETHERNET 0.420 0.400 POST i2c PASSED 0.440 0.440 POST cpu PASSED 0.920 0.920 IDE: Bus 0: OK 0.920 0.920 Device 0: Model: HITACHI_DK23DA-20 Firm: 00J2A0A1 Ser#: 12Y0MN 0.920 0.920 Type: Hard Disk 0.920 0.920 Capacity: 19077.1 MB = 18.6 GB (39070080 x 512) 2.840 2.871 Device 1: not available 2.840 2.871 SRAM: 512 kB 4.840 4.860 PS/2: No device found 5.840 5.880 Kbd: reset failed, no ACK 5.880 5.880 5.880 5.880 Type run flash_nfs to mount root filesystem over NFS 5.880 5.880 10.880 10.880 Hit any key to stop autoboot: 0 12.520 12.480 Using FEC ETHERNET device ... 13.480 13.361 Bytes transferred = 7275 (1c6b hex) 13.640 13.400 ## Booting kernel from Legacy Image at 00400000 ... 13.640 13.400 Image Name: Linux-2.6.32-rc5-01449-g2c33dca ... 13.960 13.720 ## Flattened Device Tree blob at 00600000 13.960 13.720 Booting using the fdt blob at 0x600000 14.960 13.720 Uncompressing Kernel Image ... OK 15.040 14.760 [ 0.000000] Using mpc5200-simple-platform machine description 15.040 14.760 [ 0.000000] Linux version 2.6.32-rc5-01449-g2c33dca (wd@pollux.denx.de) (gcc version 4.2.2) #1 Mon Nov 2 09:31:00 CET 2009 ...
As you can see, there are stages where the new code is a bit slower (20...40 milliseconds, which is close to the resolution of the measurement), but in this specific test we win some 280 milliseconds; if you exclude the 5 seconds boot delay that's some 2.7%. I don't think it's worth mentioning, but at least it's not worse than the old code.
I haven't tested many boards, especially not for timing. If you have additional input it will be welcome.
Best regards,
Wolfgang Denk