Re: [dpdk-dev] [PATCH] drivers/crypto: fix build with make
From: Thomas Monjalon <hidden>
Date: 2020-02-28 12:56:42
28/02/2020 13:41, Thomas Monjalon:
28/02/2020 13:24, Akhil Goyal:quoted
Hi Thomas,quoted
In the check for the version of intel-ipsec-mb library, there is a backslash in front of the #include. It is not clear why this backslash is for. It is not clear why there was no error so far. In an up-to-date ArchLinux, these errors were seen: syntax error near unexpected token `|' `grep -e "IMB_VERSION_STR" \#include <intel-ipsec-mb.h> | cut -d'"' -f2' syntax error near unexpected token `|' `grep -e "IMB_VERSION_NUM" \#include <intel-ipsec-mb.h> | cut -d' ' -f3' The makefiles are fixed by removing the backslash. Fixes: 3067c8ce77ac ("crypto/aesni_mb: fix build with custom dependency path") Fixes: 457b8e372975 ("crypto/aesni_gcm: check dependency version with make") Fixes: bf6eb2c22fd1 ("crypto/kasumi: use IPsec library") Fixes: 7c87e2d7b359 ("crypto/snow3g: use IPsec library") Fixes: 61f7c988e39e ("crypto/zuc: use IPsec library") Cc: stable@dpdk.org Signed-off-by: Thomas Monjalon <redacted> ---I see that CI is still failing after removing that backslash.Yes my patch is failing on Intel CI: http://mails.dpdk.org/archives/test-report/2020-February/118888.html I caught this issue on my machine.quoted
I believe the issue is not because of backslash. It is something else.It works on my machine. We need to find a solution which works on both environment. Maybe the cause is using make 4.3 on my machine.
I confirm the issue:
"
* WARNING: Backward-incompatibility!
Number signs (#) appearing inside a macro reference or function invocation
no longer introduce comments and should not be escaped with backslashes:
thus a call such as:
foo := $(shell echo '#')
is legal. Previously the number sign needed to be escaped, for example:
foo := $(shell echo '\#')
Now this latter will resolve to "\#". If you want to write makefiles
portable to both versions, assign the number sign to a variable:
H := \#
foo := $(shell echo '$H')
This was claimed to be fixed in 3.81, but wasn't, for some reason.
To detect this change search for 'nocomment' in the .FEATURES variable.
"
https://lists.gnu.org/archive/html/info-gnu/2020-01/msg00004.html
I will send a v2 patch