an xpath parsing error

[wqr]

from lxml.html import fromstring, tostring

html_text = """
<i><p>123</p></i>
"""

root = fromstring(html_text)
p_s = root.xpath('//i/p')

for p in p_s:
    print(tostring(p))

--------------------------

This code should output b'<p>123</p>' (if using the etree module instead of the html module, it will output correctly), but in reality, there is no output. And what's confusing is that this problem seems to only appear on the p tag under the i tag.

versions:
python: 3.10.12
lxml: 5.2.2
libxml: 2.12.6
libxslt: 1.1.39