net: ethernet: ti: am65-cpsw: fix freeing IRQ in am65_cpsw_nuss_remove_tx_chns()

[ Upstream commit 4395a44a ]

When getting the IRQ we use k3_udma_glue_tx_get_irq() which returns
negative error value on error. So not NULL check is not sufficient
to deteremine if IRQ is valid. Check that IRQ is greater then zero
to ensure it is valid.

There is no issue at probe time but at runtime user can invoke
.set_channels which results in the following call chain.
am65_cpsw_set_channels()
 am65_cpsw_nuss_update_tx_rx_chns()
  am65_cpsw_nuss_remove_tx_chns()
  am65_cpsw_nuss_init_tx_chns()

At this point if am65_cpsw_nuss_init_tx_chns() fails due to
k3_udma_glue_tx_get_irq() then tx_chn->irq will be set to a
negative value.

Then, at subsequent .set_channels with higher channel count we
will attempt to free an invalid IRQ in am65_cpsw_nuss_remove_tx_chns()
leading to a kernel warning.

The issue is present in the original commit that introduced this driver,
although there, am65_cpsw_nuss_update_tx_rx_chns() existed as
am65_cpsw_nuss_update_tx_chns().

Fixes: 93a76530 ("net: ethernet: ti: introduce am65x/j721e gigabit eth subsystem driver")
Signed-off-by: Roger Quadros <rogerq@kernel.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>