From mboxrd@z Thu Jan 1 00:00:00 1970 From: Marek Vasut Date: Sun, 25 Oct 2015 05:00:35 +0100 Subject: [U-Boot] [PATCH] mtd: do not include cfi_flash.h if no flash In-Reply-To: <1445742178-8040-1-git-send-email-thomas@wytron.com.tw> References: <1445742178-8040-1-git-send-email-thomas@wytron.com.tw> Message-ID: <201510250500.35752.marex@denx.de> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de On Sunday, October 25, 2015 at 04:02:58 AM, Thomas Chou wrote: > Do not include cfi_flash.h if CONFIG_SYS_NO_FLASH. Because > the flash_info is undefined in flash.h if CONFIG_SYS_NO_FLASH. > > Signed-off-by: Thomas Chou > --- > include/mtd/cfi_flash.h | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/include/mtd/cfi_flash.h b/include/mtd/cfi_flash.h > index 048b477..526898a 100644 > --- a/include/mtd/cfi_flash.h > +++ b/include/mtd/cfi_flash.h > @@ -8,6 +8,8 @@ > #ifndef __CFI_FLASH_H__ > #define __CFI_FLASH_H__ > > +#ifndef CONFIG_SYS_NO_FLASH > + Hi! Something tells me this is not the right way to fix things up, but I suspect this idea came from flash.h , right ? This legacy code is a terrible mess :-( I'd kinda expect that it'd be always safe to include the cfi_flash.h because cfi_flash.h would include flash.h and take care of having all the types available (not the case now). I'd also expect to have all the CFI command macros available if I include cfi_flash.h, so this new ifdef would introduce a surprising behavior for me. I wonder if we cannot find some better solution instead of adding some more ifdefs to the already horrible maze of ifdefs there. What do you think please ? > #define FLASH_CMD_CFI 0x98 > #define FLASH_CMD_READ_ID 0x90 > #define FLASH_CMD_RESET 0xff > @@ -182,4 +184,6 @@ u32 flash_read32(void *addr); > u64 flash_read64(void *addr); > #endif > > +#endif /* !CONFIG_SYS_NO_FLASH */ > + > #endif /* __CFI_FLASH_H__ */ Best regards, Marek Vasut