Re: [dpdk-dev] [dpdk-stable] [PATCH 2/4] compress/mlx5: fix constant size in QP creation

External email: Use caution opening links or attachments

-----Original Message-----
From: Thomas Monjalon <redacted>
Sent: Wednesday, June 23, 2021 9:48 AM
To: Michael Baum <redacted>
Cc: dev@dpdk.org; stable@dpdk.org; Matan Azrad <redacted>;
Raslan Darawsheh [off-list ref]; Slava Ovsiienko
[off-list ref]
Subject: Re: [dpdk-stable] [PATCH 2/4] compress/mlx5: fix constant size in
QP creation

External email: Use caution opening links or attachments


01/06/2021 09:11, Michael Baum:

quoted

The mlx5_compress_qp_setup function makes shifting to the numeric
constant 1, then sends it as a parameter to rte_calloc function.

The rte_calloc function expects to get size_t (64 bits, unsigned) and

No on 32-bit systems, size_t is 32 bits.

Thanks for the comment, I'll send v2.

quoted

instead gets a 32-bit variable, because the numeric constant size is a
32-bit.

Most of the patches of this series say "constant" where it is a variable.

quoted

In case the shift is greater than 32 the variable will lose its value
even though the function can get 64-bit argument.

Change the size of the numeric constant 1 to 64-bit.

[...]

quoted

-     opaq_buf = rte_calloc(__func__, 1u << log_ops_n,
+     opaq_buf = rte_calloc(__func__, RTE_BIT64(log_ops_n),