Hey Dan.
Couple things:
1: It might be useful to add a smatch test that looks for incompletely
initialized arrays definitions.
Neither sparse nor gcc seems to issue a warning for this possible defect.
For instance in:
drivers/staging/rtl8192e/rtl819x_HTProc.c
drivers/staging/rtl8192u/ieee80211/rtl819x_HTProc.c
There is a duplicated copy/paste array of:
u16 MCS_DATA_RATE[2][2][77] = {
{ {13, 26, 39, 52, 78, 104, 117, 130, 26, 52, 78, 104, 156, 208, 234, 260,
39, 78, 117, 234, 312, 351, 390, 52, 104, 156, 208, 312, 416, 468, 520,
0, 78, 104, 130, 117, 156, 195, 104, 130, 130, 156, 182, 182, 208, 156, 195,
195, 234, 273, 273, 312, 130, 156, 181, 156, 181, 208, 234, 208, 234, 260, 260,
286, 195, 234, 273, 234, 273, 312, 351, 312, 351, 390, 390, 429}, // Long GI, 20MHz
{14, 29, 43, 58, 87, 116, 130, 144, 29, 58, 87, 116, 173, 231, 260, 289,
43, 87, 130, 173, 260, 347, 390, 433, 58, 116, 173, 231, 347, 462, 520, 578,
0, 87, 116, 144, 130, 173, 217, 116, 144, 144, 173, 202, 202, 231, 173, 217,
217, 260, 303, 303, 347, 144, 173, 202, 173, 202, 231, 260, 231, 260, 289, 289,
318, 217, 260, 303, 260, 303, 347, 390, 347, 390, 433, 433, 477} }, // Short GI, 20MHz
{ {27, 54, 81, 108, 162, 216, 243, 270, 54, 108, 162, 216, 324, 432, 486, 540,
81, 162, 243, 324, 486, 648, 729, 810, 108, 216, 324, 432, 648, 864, 972, 1080,
12, 162, 216, 270, 243, 324, 405, 216, 270, 270, 324, 378, 378, 432, 324, 405,
405, 486, 567, 567, 648, 270, 324, 378, 324, 378, 432, 486, 432, 486, 540, 540,
594, 405, 486, 567, 486, 567, 648, 729, 648, 729, 810, 810, 891}, // Long GI, 40MHz
{30, 60, 90, 120, 180, 240, 270, 300, 60, 120, 180, 240, 360, 480, 540, 600,
90, 180, 270, 360, 540, 720, 810, 900, 120, 240, 360, 480, 720, 960, 1080, 1200,
13, 180, 240, 300, 270, 360, 450, 240, 300, 300, 360, 420, 420, 480, 360, 450,
450, 540, 630, 630, 720, 300, 360, 420, 360, 420, 480, 540, 480, 540, 600, 600,
660, 450, 540, 630, 540, 630, 720, 810, 720, 810, 900, 900, 990} } // Short GI, 40MHz
where it appears that the first 77 entries are one short using only
76 initializations rather than 77.
(Perhaps the 2nd initialization line of
39, 78, 117, 234, 312, 351, 390, 52, 104, 156, 208, 312, 416, 468, 520,
should be
39, 78, 117, 234, 273, 312, 351, 390, 52, 104, 156, 208, 312, 416, 468, 520,
I don't know and don't want to know the code, but it seems odd at best)
I haven't inspected smatch code for the capability, but I didn't find
any similar output message in the code and there isn't a smatch
mailing list archive to search.
2: lore.kernel.org does not have an archive of the smatch mailing list
You might ask for one to be created at bugzilla.kernel.org and give
them a copy or a pointer to an archive if you have one.
